Florida State University Libraries

Electronic Theses, Treatises and Dissertations The Graduate School

2012 Effects of Scale Insects on Forest Dynamics in Fragmented Tropical Montane Oak Forests of Veracruz, Mexico Heather A. Gamper

Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected] THE FLORIDA STATE UNIVERSITY

COLLEGE OF SOCIAL SCIENCE AND PUBLIC POLICY

EFFECTS OF SCALE INSECTS ON FOREST DYNAMICS IN FRAGMENTED

TROPICAL MONTANE OAK FORESTS OF VERACRUZ, MEXICO

By

HEATHER A. GAMPER

A Dissertation submitted to the Department of Geography in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Degree Awarded: Spring Semester, 2012 Heather A. Gamper defended this dissertation on March 30, 2012. The members of the supervisory committee were:

Anthony J. Stallins Professor Co-Directing Dissertation

James Elsner Professor Co-Directing Dissertation

Brian Inouye University Representative

Xiaojun Yang Committee Member

Victor Mesev Committee Member

The Graduate School has verified and approved the above-named committee members, and certifies that the dissertation has been approved in accordance with the university requirements.

ii to my Grandmother - Virginia Marie Gardella (1919 -2010)

iii ACKNOWLEDGMENTS

This work is not mine alone, for its success weighs on many others. The origination of studying scale insects began in the countryside of the Dominican Republic, where I was a part of a group of ornithologists who discovered and cooperatively described a unique phenomenon of birds feeding from scale insect secretions. I was the only one in this group passionate, or perhaps foolish enough, to focus my future research efforts on such an odd topic. I was able to search for other occurrences of this phenomenon after being accepted as a Biology Master’s degree student by Dr. Suzanne Koptur at Florida International University. Much of my understanding of plant-animal interactions was cultivated by Suzanne, a wise and motherly mentor. Suzanne gave me the confidence to explore the tropics in pursuit of scale insects which conceptually began my life as an independent scientist. I am thankful in this intricate network of life, our webs collided. At the same time I collided with my life’s best friend and husband, Andres, who has supported me and has been by my side from the moment we met. Living in a foreign country could be uncomfortable and intimidating, but it rarely was because of two amazing women: Mariana Cuautle Arenas and Cecilia Castelazo. They welcomed me into their homes for many years, fed me the most delicious food I have ever eaten and incorporated me into their group of friends. Night after night, after working in the forest, I came home to a place where I have never lived so contently within the glow, laughter and pulse of Mexico. At the Institute of Ecology in Xalapa Mexico, I was fortunate to find two tremendous mentors: Dr. Guadalupe Williams-Linera, and Dr. Jose ’Pepe’ Garcia Franco who provided invaluable logistical support. Throughout the years this work was completed, there were many people who assisted me in the field: Victor Ruiz Mandoval, Miriam Ferrer, Mariana Cuautle, Cecilia Castelazo, Socorro Jiminez, Christopher Brown, and Andres Plata. I owe the decision to pursuing my Doctoral Degree at FSU, to Dr. Brian Inouye in the Biology Department. Brian gave me wise guidance for studying this ecological interaction and I am grateful to have been given the opportunity to work with him and look forward to working with him in the future. Within the geography department, I am appreciative of the teaching, advising, and guidance of Dr. Elsner, Dr. Yang, Dr. Mesev, Dr. Tschinkel and Dr. Steinberg. I am particularly thankful to Dr. Anthony ’Tony’ Stallins for his patience with me and care in crafting me into a geographer. I am especially thankful for his reading and editing of my work in a fashion that exemplifies his capabilities as an advisor. During the analysis and writing stage of my dissertation, I mentored an undergraduate student, Brandy Safell. In addition to completing an independent research project in den-

iv drochronology, Brandy was extremely eager and helpful in learning and assisting with my dissertation research. Brandy helped create maps and participated in statistical analyses that we jointly presented at meetings. Brandy is a brilliant student and I am fortunate to have had the opportunity to work with her. Many beekeepers deserve thanks for increasing my knowledge of the subject and making the beekeeping chapter more pertinent. In particular, I thank Kelly Watson for sharing with me her research at FSU and introducing me to the bees. I also owe much thanks to Donald Smiley in Wewahitchka, Florida for making me a beekeeper. If it were not for a strong network of friends and family, this work may have never been completed. Kim McClellan allowed me to ease back into teaching shortly after my daughter Celeste Wren was born by accepting her with loving arms into her office while I gave weekly lectures. I must thank my parents Robert and Virginia Gamper and my husband Andres Plata. Andres has kept me going and has assisted me with this work in both intellectual and spiritual ways. It is with his support that I was capable of simultaneously pouring every bit of energy I had into both my love for him and Celeste and my passion for science. Thank you.

v TABLE OF CONTENTS

ListofTables...... ix ListofFigures ...... x Abstract...... xiv

1 Introduction 1 1.1 Scale insects in tropical montane cloud forest of Veracruz, Mexico ...... 1 1.2 Studysite...... 2 1.3 Dissertation organization ...... 6

2 Honeydew Foraging by Birds in Tropical Montane Forests and Pastures of Mexico 8 2.1 Abstract...... 8 2.2 Introduction...... 8 2.3 Methods...... 10 2.3.1 Study area ...... 10 2.3.2 Bird foraging observations in oak trees ...... 10 2.3.3 Statistical analysis ...... 12 2.4 Results...... 12 2.4.1 Bird foraging ...... 12 2.4.2 Interspecific interactions and resource defense ...... 14 2.4.3 Comparison of forest and pasture ...... 14 2.5 Discussion...... 14

3 Alteration of forest structure modifies the distribution of scale insect, Stigmacoccus garmilleri, in Mexican tropical montane cloud forests 19 3.1 Abstract...... 19 3.2 Introduction...... 20 3.3 Methods...... 24 3.3.1 Studysite...... 24 3.3.2 Scale insect density measures ...... 24 3.3.3 Anal filament and honeydew drop measures at observation trees . . 26 3.3.4 Anal filament growth rate at observation trees ...... 27 3.4 Results...... 27 3.4.1 Anal filament and honeydew drop measurements ...... 27 3.4.2 Anal filament growth ...... 31

vi 3.5 Discussion...... 34 3.5.1 Scale insect distribution ...... 34 3.5.2 Honeydew volume/concentration ...... 35 3.5.3 Anal filaments ...... 35 3.6 Conclusions...... 36

4 Explaining the relationship between chronic scale insect herbivory and tree growth in fragmented forests of Veracruz, Mexico 38 4.1 Abstract...... 38 4.2 Introduction...... 39 4.2.1 Forest change and scale insect distribution ...... 40 4.3 Methods...... 41 4.3.1 Dendrochronology in the tropics ...... 41 4.3.2 Study area ...... 43 4.3.3 Field data collection ...... 43 4.3.4 Sample preparation and measurements ...... 45 4.3.5 Pithflecks...... 46 4.4 Results...... 46 4.5 Discussion...... 55

5 A socioecological critique of beekeeping as mechanism to offset deforesta- tion: the importance of context for honey production in rural Veracruz, Mexico 65 5.1 Abstract...... 65 5.2 Introduction...... 66 5.3 Ecological Knowledge ...... 67 5.3.1 Land-use change and landscape configuration ...... 68 5.3.2 Non-native pollinators ...... 69 5.3.3 Selectingabeespecies...... 70 5.3.4 Access to floral resources ...... 70 5.3.5 Mobility of Beekeeping ...... 71 5.4 Socioecological context ...... 71 5.4.1 Social networks ...... 71 5.4.2 Addressing economic opportunity ...... 72 5.4.3 Gender roles - women participants ...... 73 5.4.4 Negotiating access to forage ...... 73 5.4.5 Pests, pathogens and africanized bees ...... 74 5.4.6 Challenges to marketing honey ...... 74 5.4.7 Non-floral resources ...... 75 5.5 Casestudyexample: VeracruzMexico ...... 78 5.6 Conclusions...... 82

6 Conclusions 83

vii A Maps illustrating the distribution of scale insect, Stigmacoccus garmilleri in tropical montane forests of Chiconquiaco, Mexico 86 A.1 Methods...... 86 A.1.1 GPS data collection ...... 86 A.1.2 Geovisualization ...... 86 Bibliography ...... 95 Biographical Sketch ...... 115

viii LIST OF TABLES

2.1 Bird visits to scale insect honeydew on oak trees in tropical montane forest and pasture areas of Chiconquiaco, Veracruz, Mexico...... 13

2.2 Summary of birds and their feeding behavior on oak trees in tropical montane forest and pasture areas of Chiconquiaco, Veracruz, Mexico...... 16

3.1 Means, standard deviations, minimum and maximum values for honeydew- drop volume, sugar concentrations, and anal-tube lengths for scale insects (Stigmacoccus garmilleri) on oaks in forest (n = 756) and pasture habitats (n = 762) of Chiconquiaco, Mexico...... 30

3.2 Spearman’s correlation coefficients were calculated for the predictive strength of the recorded variables (temperature, relative humidity, drop volume, sugar concentration, and anal-tube length)...... 31

4.1 Chronology statistics for all species in all habitats. Series intercorrelation is a statistic that demonstrates the correlation of each core to the master chronology and is a measure of stand-level signal. Mean sensitivity measures the year-to-year variability and is a measure of sensitivity. The chronology length reports the length of the oldest core...... 48

4.2 Analysis of variance results of tree growth as a function of scale insect density, tree size and location (forest edge or interior)...... 52

4.3 Analysis of covariance results of tree growth as a function of tree size and scale insectdensityasthecovariate...... 53

ix LIST OF FIGURES

1.1 Research location within tropical montane forest near the town of Chicon- quiaco, in the state of Veracruz, Mexico ...... 3

1.3 Unsupervised remote sensing classification of 1989 Landsat imagery surround- ing the study area describes 8 land cover/land use categories (Steep forest, dense forest, forest, fragmented forest, pasture areas with remnant forest trees, pasture, pasture mixed with soil, pasture mixed with rural development). . . 4

1.4 Unsupervised remote sensing classification of 2000 Landsat Enhanced The- matic Mapper Plus (ETM+) imagery describes 8 land cover/land use cate- gories (steep forest, dense forest, forest, fragmented forest, pasture areas with remnant forest trees, pasture, pasture mixed with soil, pasture mixed with ruraldevelopment)...... 5

2.1 Morphology of scale insects producing honeydew on oak trees in tropical mon- tane cloud forest of Chiconquiaco, Veracruz, Mexico. Anal filaments of these insects are visible excreting the sugary, honeydew waste. (a). Individual scale insect (Stigmacoccus garmilleri) found on oak (Quercus spp.) in late feeding instar stage. Excrement of this insect (honeydew) is visible at the end of the anal filament. (b). Colony of scale insects (Stigmacoccus garmilleri) on the trunk of oak (Quercus spp.)...... 11

2.2 Diagrammatic representation of interactions among birds utilizing honeydew on trees harbouring scale insects during 160 1-h observations on observation trees in Chiconquiaco, Veracruz, Mexico...... 15

3.1 Hair-like anal filaments from scale insect Stigmacoccus garmilleri and drops of honeydew secreted from their ends (Chiconquiaco, Veracruz, Mexico).Scale insects are capable of reaching high densities on oak tree trunks and branches. 21

3.2 Winged adult male (Stigmacoccus garmilleri) mating with soft-bodied adult female. For approximately three weeks during their life cycle adults are found mating on the trunks of host trees (Quercus spp.) in tropical montane forests of central Veracruz, Mexico...... 22

x 3.3 Invertebrate species are known to feed on honeydew produced by scale insect Stigmacoccus garmilleri in tropical montane forests of central Veracruz, Mex- ico. (a) Mite (Anystis sp.) found foraging on a honeydew droplet and (b) Vespid wasp in flight consuming honeydew...... 23

3.4 The study was conducted in tropical montane forest near the town of Chicon- quiaco, in the state of Veracruz, Mexico ...... 25

3.5 Mean estimates (95% confidence interval, standard deviation) of scale-insect density per 400 cm2 on trees at different heights in forest habitat...... 28

3.6 Mean estimates (95% confidence interval, standard deviation) of scale-insect density per 400 cm2 on trees of different size classes within forested habitat. Small = 5-28.6 cm, medium = 29-46.3 cm, large = 48.9-83 cm diameters at breastheight ...... 29

3.7 Honeydew sugar concentration plotted against temperature recorded at time of measurement. Data were collected from 20 trees in forest habitat and 20 trees located in pasture habitat in Chiconquiaco, Mexico...... 32

3.8 Honeydew sugar concentration plotted against relative humidity recorded at time of measurement. Data were collected from 20 trees in forest habitat and 20 trees located in pasture habitat in Chiconquiaco, Mexico...... 33

4.1 How tree growth rates may be influenced by the light environment and scale insect density in tropical montane cloud forests of Chiconquiaco, Mexico. . . 42

4.2 Climograph illustrating average monthly precipitation and temperature (1996- 2003) in Naolinco de Victoria, Veracruz (10 km from study site)...... 43

4.3 The two forest areas from which tree core samples were collected:(A) Forest area PP showing a heavily grazed understory, with little to no new tree estab- lishment in the understory or forest edge, (B) Forest area BB showing a more diverse range of age and size classes and less grazing present in the understory. 44

4.4 Tree ring boundaries formed in a ring porous oak species, Quercus laurina. Olderringsaretotheleft ...... 47

4.5 The time span and series overlap of each tree core sample collected in 2008 from forest area BB in Chiconquiaco, Mexico ...... 49

4.6 The time span and series overlap of each tree core sample collected in 2008 from forest area PP in Chiconquiaco, Mexico ...... 50

4.7 The standard chronology is shown for the tree cores collected at site PP using raw ring width data. A smoothing spline highlights low-frequency variability and the sample depth (number of tree cores used in the chronology for that time period) is plotted on the right-hand y-axis with a dotted line ...... 52

xi 4.8 The standard chronology is shown for the tree cores collected at site BB using raw ring width data. A smoothing spline highlights low-frequency variability and the sample depth (number of tree cores used in the chronology for that time period) is plotted on the right-hand y-axis with a dotted line ...... 53

4.9 Detrended chronology for the tree cores collected at site BB using basal area index values. A smoothing spline highlights low-frequency variability and the sample depth (number of tree cores used in the chronology for that time period) is plotted on the right-hand y-axis with a dotted line ...... 54

4.10 Detrended chronology for the tree cores collected at site PP using basal area index values. A smoothing spline highlights low-frequency variability and the sample depth (number of tree cores used in the chronology for that time period) is plotted on the right-hand y-axis with a dotted line ...... 55

4.11 Mean basal area growth plotted as a function of tree diameter for all sample treesinChiconquiaco,Mexico ...... 56

4.12 Tree locations at both study sites (BB, PP) mapped with increasing symbol size as a function of increasing tree diameter and with color categories indi- cating annual mean growth ...... 57

4.13 Locally weighted regression of tree growth as a function of scale insect density and tree size classes (small diameter (5.2-15 cm), medium diameter (15.5-39 cm), and large diameter (> 39 cm)) using the locally weighted scatterplot smoothing function. Data were combined from forest site BB and PP. . . . . 58

4.14 Locally weighted regression of tree growth as a function of scale insect density and location (forest edge, forest interior) and location and size (small trees on the forest edge) using the locally weighted scatterplot smoothing function. Data were combined from forest site BB and PP...... 59

4.15 Small areas of wood resembling tree pith (’pith flecks’) are caused by damage to tree cambium by the feeding of scale insect, Stigmacoccus garmilleri. These pith flecks are visible and highlighted here within prepared tree cores from Quercus laurina in Chiconquaico, Mexico...... 60

4.16 Individual years pith flecks were observed (Forest area BB, tree 22) plotted upon yearly growth increment values for the lifespan of the tree...... 61

4.17 Map illustrating the relationship between tree core sample location, scale insect density, and numerical occurence of pith flecks within core samples for each tree. 62

4.18 Degree of clustering of trees with observed pith flecks plotted by Ripley’s K function tested against a confidence envelope for complete spatial randomness (CSR) created by performing simulations or random permutations...... 63

xii 5.1 Market display of the variety of honey colors and types produced by beekeepers from the state of Veracruz, Mexico ...... 76

A.1 Distribution and relative position of the individal forest plots where scale insect distribution mapping on individual trees was conducted in 2007 Chiconquiaco, Veracruz,Mexico...... 87

A.2 Trees mapped within individual forest plots (fores site BB; plot 1 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Attribute data such as tree crown size and tree trunk diameter at breast height (DBH) also aided in conceptualizing forest structure. . . . . 88

A.3 Trees mapped within individual forest plots (fores site BB; plot 2 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Attribute data such as tree crown size and tree trunk diameter at breast height (DBH) also aided in conceptualizing forest structure. . . . . 89

A.4 Trees mapped within individual forest plots (fores site BB; plot 1 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Data collected at each individual tree on scale insect density at varying tree heights (lower 1/3, middle 1/3, and upper 1/3) were used to createscaleinsectdensitymaps...... 90

A.5 Trees mapped within individual forest plots (fores site BB; plot 2 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Data collected at each individual tree on scale insect density at varying tree heights (lower 1/3, middle 1/3, and upper 1/3) were used to createscaleinsectdensitymaps...... 91

A.6 Trees mapped within individual forest plots (fores site BB; plot 5 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Data collected at each individual tree on scale insect density at varying tree heights (lower 1/3, middle 1/3, and upper 1/3) were used to createscaleinsectdensitymaps...... 92

A.7 Trees mapped within individual forest plots (fores site PP; plot 1 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Data collected at each individual tree on scale insect density at varying tree heights (lower 1/3, middle 1/3, and upper 1/3) were used to createscaleinsectdensitymaps...... 93

A.8 Trees mapped within individual forest plots (fores site PP; plot 2 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Data collected at each individual tree on scale insect density at varying tree heights (lower 1/3, middle 1/3, and upper 1/3) were used to createscaleinsectdensitymaps...... 94

xiii ABSTRACT

The montane cloud forests of northeastern Mexico have a high concentration of endemism, and are increasingly vulnerable to climate change, deforestation, and habitat fragmentation. Ninety percent of the original montane cloud forest in Mexico has been lost. As habitat fragmentation becomes more pervasive throughout the world, our understanding of forest fragmentation has also grown more sensitive to context. The species interactions initiated with fragmentation, and post-fragmentation disturbance regimes may magnify the impacts of fragmentation. This dissertation research seeks to develop a better understanding of post-fragmentation pattern and processes in the montane oak cloud forests of northeastern Mexico. Central to this goal is the documentation of a forest change occurring within these frag- mented forests. An endemic scale insect, Stigmacoccus garmilleri, which typically occurs in low densities in the upper canopy of intact oak forests, has expanded throughout for- est fragments within the oak forests of Chiconquiaco, Mexico. Observations indicate that S. garmilleri is common within a 10km 2 area around the research study site, and likely extends further. This research focuses on how the high population densities of S. garmilleri may augment local diversity in fragmented habitat by providing a food source upon which other species depend. Of critical importance is how these scale insects, at the high densities that augment diversity, impact their host trees. The majority of scale insects are considered pests in agricultural settings. How the host oaks respond to scale determines not only the diversity of these forest fragments, but also their future persistence. As part of this dissertation research, the potential for these fragmented forested areas to support beekeeping and lessen the dependence upon the agricultural land uses that are driving forest fragmentation are discussed. The sociecological context of deforestation and the promotion of beekeeping for the township of Chiconquiaco lends itself to critical examination of apicultural development. This case study in Veracruz state serves as a template for a more adaptive framework for implementing forest conservation beekeeping projects.

xiv CHAPTER 1

INTRODUCTION

1.1 Scale insects in tropical montane cloud forest of Veracruz, Mexico

The interaction between species has a considerable influence on diversity. Most studies of species interactions focus on predation, competition, parasitism and mutualism. How- ever, species can influence the diversity of their associated communities through their role as ecosystem engineers (Bruno, 2000; Stachowicz, 2001; Bruno et al., 2003). Ecosystem engineers facilitate the occurrence of other species by creating, maintaining, or modifying habitats (Jones et al., 1994, 1997; Lawton, 1994). Ecosystem engineers alter their physical surroundings or change the flow of resources, thereby creating or modifying habitats (Crain and Bertness, 2006). The concept of ecosystem engineers is now largely accepted by the ecological community (e.g., Wright and Jones 2006; Hastings et al., 2007). Nevertheless, most studies of ecosystem engineering invoke only equilibrium concepts of stability. Less is known about the tradeoffs for engineer species (Bouma et al., 2005). These simplifica- tions may be a consequence of the association of ecosystem engineering with strong Gaian concepts of homeostatis and self-regulation. Although it is recognized that environmental change may introduce conditions that initiate ecosystem engineering, the engineering effects that follow may not necessarily act in a direction to reestablish initial conditions (Folke et al., 2004; van Nes and Scheffer, 2004; Stallins, 2006). Critics of the ecosystem engineering concept note that all organisms affect and are effected by their physical environment (Reichman and Seabloom, 2002). Supporters of the ecosystem engineering concept have stressed that its utility depends on understanding when the modification of the environment needs to be elucidated upon rather than only considered as a part of the direct interactions between species. Explicit inclusion of the concept is required when the structural changes to the environment persist on time scales longer than the individual lifetime of the organism (Hastings et al., 2007). This dissertation frames scale insect Stigmacoccus garmilleri as an ecosystem engineer in these fragmented forests not only because the thick organic substrate it creates may persist long after the life span of individual scale insects, but also because of the modification of food resources by S. garmilleri when it occurs at high densities. The presence of scale insects in these forest fragments results in a very different habitat than the intact surrounding forests. Scale insects are known for their honeydew secretions, the relationships with ants that

1 feed on them, and their propensity to increase local diversity (Buckley, 1987; Bach, 1991). What is unique about the development of this engineering effect on diversity in this disser- tation is that it may be linked to historic fragmentation. Fragmentation in tropical forests results in warmer temperatures and wider temperature ranges (Laurance, 2004). Observa- tions conducted over visits to the study site suggest that the current distribution of scale insects and their sooty mold substrate have a dependence upon edge microclimates created by fragmentation. Densities of S. garmilleri appear to be greatest on trees located along forest edges or isolated in pasture areas. In adjacent closed forests, scale insects are confined to the canopy. A greater flux in daily and seasonal temperatures may benefit scale insect feeding because temperature oscillations can enhance sap flow in trees. Other studies have shown that forest insect pests can increase in response to human-driven fragmentation and the spatial patterns of environmental conditions that result (Coulson et al., 1999; Johnson et al., 2006; Moreau et al., 2006; Grilli and Bruno, 2007). This dissertation postulates that there may exist a tradeoff between the diversity- enhancing effects of scale insects, and the role in may play as a forest pest. Although scale insects can enhance diversity, by feeding on the phloem of their host oaks they may exert a negative influence on tree growth and reproduction. Taken out of its continuous forest type, S.garmilleri may have short term benefits for overall diversity, but detrimental effects on host tree regeneration. This dissertation explores the hypothesis that if S. garmilleri slows regeneration, its engineering effects may be self-limiting. In this positive feedback (Wil- son and Agnew, 1992), declining regeneration may lead to warmer edge conditions, more infestation and more engineered local diversity, but an overall intrinsically-driven lock-in of forest fragmentation and decline. In this perspective, this dissertation contributes to a more nuanced understanding of ecosystem engineering, and how it can drive a contextual and spatially explicit response to environmental change (Wright and Jones 2006)

1.2 Study site

The montane cloud forests of northeastern Mexico have a high concentration of en- demism, and are increasingly vulnerable to climate change, deforestation, and habitat frag- mentation (Markham, 1998; Bubb et al., 2004). Ninety percent of the original montane cloud forest in Mexico has been lost (Ramirez-Marcial et al., 2001). As habitat fragmenta- tion becomes more pervasive throughout the world, our understanding of forest fragmenta- tion has also grown more sensitive to context. The types of edge, the species interactions initiated with fragmentation, and post-fragmentation disturbance regimes may magnify the impacts of fragmentation (Laurance, 2002; Ewers and Didham, 2006; Malanson et al., 2007). This dissertation research seeks to develop a better understanding of post-fragmentation pattern and processes in the montane oak cloud forests of northeastern Mexico. The region under study is the low volcanic ranges of the Sierra de Chiconquiaco that interrupt the coastal plain of Veracruz state, Mexico. These mountains face the Gulf of Mexico, and receive torrential summer rains at lower elevations on ocean-facing slopes while subhumid rain shadows develop on leeward slopes (Pool, 2006). The area is characterized by three climatic seasons, although heavy fog is common on most days: somewhat dry and cool from October through March, dry and warm from April through May, and wet and

2 Figure 1.1: Research location within tropical montane forest near the town of Chiconquiaco, in the state of Veracruz, Mexico

warm from June through September (Williams-Linera et al., 2000). The specific study site is located within the township of Chiconquiaco at roughly 2000 m elevation 1.1. The current municipal area of Chiconquiaco is considered rural, with approximately 2900 inhabitants and of these inhabitants 26% are indigenous. Dairy cattle grazing is the most dominant land-use practice in the remaining montane cloud forest of Chiconquiaco (INEGI, 2005). Tropical montane forests within the region are discontinuously distributed ??. Re- maining forests commonly form an abrupt boundary with adjacent grazed grassland, small agricultural fields, and clearings for houses and community structures. Forest fragments have low edge to interior area ratios and are comprised of a canopy of hybridizing assem- blages of Q. laurina, Q. germana, Q. salicifolia, Q. corrugata, Q. affinis, Q. crassifolia, and Q. xalapensis. All of these oaks are apparently capable of hosting the scale insect S. garmilleri. Three oaks are predominant in the study area: Q. laurina, Q. affinis, and Q. crassifolia. Pastures consist of small clumps and isolated individuals of Quercus spp. The montane cloud forests of northeastern Mexico have a high concentration of en-

3 1989

Steep forest Dense forest Forest Fragmented forest

Pasture with trees Pasture Pasture/soil Pasture/rural development

Figure 1.3: Unsupervised remote sensing classification of 1989 Landsat imagery surrounding the study area describes 8 land cover/land use categories (Steep forest, dense forest, forest, fragmented forest, pasture areas with remnant forest trees, pasture, pasture mixed with soil, pasture mixed with rural development).

4 2000

Steep forest Dense forest

Forest Fragmented forest

Pasture with trees Pasture Pasture/soil Pasture/rural development

Figure 1.4: Unsupervised remote sensing classification of 2000 Landsat Enhanced Thematic Mapper Plus (ETM+) imagery describes 8 land cover/land use cat- egories (steep forest, dense forest, forest, fragmented forest, pasture areas with remnant forest trees, pasture, pasture mixed with soil, pasture mixed with rural development).

5 demism and are increasingly vulnerable to climate change, deforestation, and habitat frag- mentation (Markham 1998; Bubb et al. 2004). Ninety percent of the original montane cloud forest in Mexico has been lost (Ramirez-Marcial et al. 2001). Remaining forests are discontinuously distributed and commonly form an abrupt boundary with adjacent grazed grassland, small agricultural fields, and clearings for houses and community structures. Classification of Landsat satellite images (TM 1989, ETM+ 2000, (unsupervised classsifi- cation, 8 classes)) was used to estimate gains and losses over this 11 year time period in the area surrounding Chiconquiaco 1.3, 1.4. The following gains (+) and losses (-) were calculated for the surrounding study area: steep forest (- 30.73%), dense forest (- 26.47%), forest (- 24.35%), fragmented forest (+59.74%), pasture with scattered trees (- 77.23%), pasture (+ 55.51%), pasture/soil (+ 55.52%), and rural development (+ 96.6%). A general trend was observed for the conversion of larger more intact areas of tropical montane forest to other land cover/land use categories. A decrease of pasture areas with scattered trees was observed due to the removal of these remnant trees. Overall a general increase of pasture and rural development was observed. As habitat fragmentation becomes more pervasive throughout the world, our under- standing of forest fragmentation has also grown more sensitive to context. The species interactions initiated with fragmentation and post-fragmentation disturbance regimes may magnify the impacts of fragmentation (Laurance 2002; Ewers and Didham 2006; Malanson et al. 2007). Forest fragments in this study are biotic representatives of formerly intact montane cloud forests. Understanding their persistence and interactions with existing land use will permit us to broaden our impacts by contributing to dialogues on their management (Chazdon, 2003).

1.3 Dissertation organization

Tropical montane cloud forests are a rare forest type, contain high levels of endemic species and are important in the ecosystem services they provide, most notably for their potential for water capture. This research is a study of how land-use changes surround- ing these forests can alter the dynamics of remaining tropical montane forest patches in Veracruz, Mexico. This research also engages in broader perspectives surrounding the chal- lenges of understanding how insect densities alter forest processes. Chapter two of this dissertation describes the importance of honeydew producing scale insects for other organisms in the focal study area. In summary, it describes the reliance of many bird species on this rich carbohydrate resource and outlines how the use (consumption and even defense) of scale insect honeydew by the bird community in forest trees differs from how it is used in isolated pasture trees. This chapter is a manuscript that was published in the Journal of Tropical Ecology in 2010 (26: 335-341). The use of ’we’ and ’our’ in this chapter and in other chapters of the dissertation refers to myself and additional co-authors who were involved or will be included on manuscripts in submission. Chapter three reveals how land-use change has altered the distribution of this important species of scale insect, Stigmacoccus garmileri. Scale insect density, honeydew volume, and sugar concentration were surveyed throughout a continuous landscape that included both patches of forest and scattered pasture trees. Results obtained from this data collection

6 describe the increases in scale insect infestation of trees with forest disturbance. These results have been published in the Journal of Insect Science ((2011) Vol 11: Article 120). Chapter four investigates the physiological effects on the host tree from this observed increase in scale insect density within disturbed forest areas. To undertake this research question I adopted a dendrochronological (tree-ring study) approach. Tree cores were col- lected from the sample trees with varying: densities of scale insects, distance from forest edge, and tree diameter size classes. This sampling strategy allowed for the interpretation of how these various factors affect oak tree growth. Information on tree growth from small trees regenerating on the forest edge with populations of scale insects aid in our under- standing of how changes to biotic interactions (such as these insect-plant interactions) from human land use change can in turn alter processes such as forest regeneration and ecological restoration efforts in the region. In the face of substantial forest loss and difficulties of forest recovery in this region, the research in chapter five challenges those involved in bettering rural livelihoods to rethink the approach of implementing conservation strategies that promote non-timber forest products particularly the production of honey. Specifically this study demonstrates the importance of incorporating local knowledge into productive forest conservation strategies that involve rural beekeeping. Scale insect honeydew is discussed as a case study for producing honey- dew honey with the potential in development of a sucessful rural beekeeping program in Chiconquiaco, Mexico. textcomp amsmath

7 CHAPTER 2

HONEYDEW FORAGING BY BIRDS IN TROPICAL MONTANE FORESTS AND PASTURES OF MEXICO

2.1 Abstract

Honeydew, a sugar solution produced as waste by phloem-feeding insects, is a prized food for many species of ants. A honeydew-producing scale insect, Stigmacoccus garmilleri Hempel (family Margarodidae), is associated with oak trees (Quercus spp.) in highland forests of Mexico. Although feeding by ants on scale-insect honeydew is more frequently documented in the literature, the honeydew produced by feeding instars of S. garmilleri is sufficient to provide nourishment for birds. This study elucidates bird use of honeydew in the tropical montane forests near Chiconquiaco, Veracruz, Mexico, and uncovers patterns in honeydew foraging. Over a two month time period, forty trees harboring scale insects, located in both forest and pasture areas, were intensely studied (160 hours of bird forag- ing observations along with quantitative measurements of honeydew production). Fifteen resident bird species and 18 migrant species were observed visiting observation trees. Ap- proximately 72% of the resident bird species and 83% of the migrant bird species observed were recorded to forage on scale-insect honeydew. Audubon’s warbler (Dendroica coronata auduboni) was the most active consumer and defender of the resource. Of 118 aggressive chases observed, only 9.65% occurred in forest observation trees, and 90.35% in pasture trees. When forests are converted to more open agricultural habitats, the importance of isolated resources may increase, as is supported by the preferential defense and territorial patrolling of scale-insect honeydew by Audubon’s warbler in scattered pasture trees.

2.2 Introduction

Honeydew is a sugary excretion of phloem-feeding insects. Phloem sap passes through the digestive system of the insect largely unchanged (Way 1963). It contains large amounts of carbohydrates and trace amounts of amino acids and can provide an important food source. Honeydew-producing insects tend to excrete copious honeydew, live in groups, and are typically sedentary or semi-sedentary (Williams and Williams 1980). Scale insects (Coc- coidea) are well known for their honeydew secretions and relationships with ants that feed

8 on them (reviewed by Bach 1991, Buckley 1987, Way 1963). Less frequently documented is the use of honeydew by birds. Several examples have been described from New Zealand (Beggs 2001, Gaze and Clout 1983, Moller and Tilley 1989, Murphy and Kelly 2001, 2003), Australia (Loyn et al. 1983, Paton 1980, Woinarski 1984), Costa Rica (Jirn and Salas 1975), Colombia (Koster and Stoewesand 1973), Brazil (Reichholf and Reichholf 1973), Mexico (Edwards 1982, Greenberg et al. 1993, Hodgson et al. 2007), and the Dominican Republic (Latta et al. 2001). A honeydew-producing scale insect, Stigmacoccus garmilleri Hempel, is associated with oak trees (Quercus spp.) in highland forests of Chiapas, Mexico (Greenberg et al. 1993). Quercus spp. in forests of Veracruz, Mexico, the focal area of this study, harbour the same species (Hodgson et al. 2007) 3.1. Ants, the usual consumers of scale insect honeydew were only occasionally observed foraging on honeydew, while migrant and resident birds were commonly found foraging on honeydew (Hodgson et al. 2007). We occasionally observed other Hymenoptera (bees and wasps), in addition to Diptera (flies) and Acaridae (mites). Migratory birds do not commonly engage in interspecific aggression in wintering habitat (Greenberg et al. 1996), but Greenberg et al. (1993) found that where such aggression occurs, it is most common when birds are using resource-rich patches such as fruit or nectar. Several studies have documented bird defense of honeydew (Greenberg et al. 1993, Latta and Faaborg 2002, Latta et al. 2001, Paton 1980, Woinarski 1984). Honeyeaters (Lichenostomus sp. and Manorina sp.) defend dense patches of lerp (sugary exudates from psyllids) and in extreme cases exclude all other bird species (Woinarski 1984). Paton (1980) observed honeyeaters defending trees from other species as well as from conspecifics. In addition to intraspecific and interspecific interactions, intersexual dominance interactions within species have been noted in defense of honeydew (Greenberg et al. 1993, Latta and Faaborg 2002, Latta et al. 2001). Along with the type and richness of the food involved, habitat structure may play an important role in the defensibility of a resource. Single trees in open pasture may be more defensible than a structurally diverse forest with a greater abundance of birds (Orians and Willson 1964). Greenberg et al. (1996) discovered that yellow warblers (Dendroica petechia) invest large amounts of energy in displacing birds entering the small number of trees located within their pasture territories. In the study area in Veracruz, Mexico, the Quercus spp. with dense colonies of scale insects are found both in forest habitat and as isolated trees in cow pastures. The bird community is diverse; 55 bird species were observed visiting oak trees in the vicinity of the observational trees (Gamper pers. obs.). During winter, the area is home to many Neotropical migrant birds. In addition, many species endemic to Mexico are found in oak- dominated forests (Watson 2003). This study aimed to describe bird use of honeydew in the study area and uncover potential patterns and influential factors to honeydew foraging. This study also compares bird defense of the honeydew resource in forest and pasture areas. We hypothesized that aggressive interactions among birds would be more frequent in honeydew trees isolated in the pasture areas than in those within forest. Differences in honeydew resource distribution varies between forest and pasture habitat and is the subject of subsequent investigation.

9 2.3 Methods

2.3.1 Study area The study reported here was conducted near the town of Chiconquiaco, in the state of Veracruz, Mexico. The area has three seasons: moderately dry and cool from October through March, dry and warm from April through May, and wet and warm from June through September (mean annual temperature is 15.2➦C; total mean annual precipitation is 1532 mm) (Williams-Linera et al. 2000). The area is covered by heavy fog on most days and humidity levels remain high even during the relatively dry months. The elevation is approximately 2000 m, and the habitat is a mosaic of mature forest patches, cattle pastures, and small cornfields. The observational trees were located in two forest patches (each ca. 25 ha) and two pasture areas (each ca. 35 ha), all on west-facing slopes. Forest areas had closed canopies dominated by Quercus spp., and pasture areas were open and included a few, large, scattered Quercus spp. individuals. Several species of oak have been identified from the study area (Q. laurina Bonpl., Q. germana Schltdl. and Cham., Q. salicifolia Ne., Q. corrugata Hook., Q. affinis Schweid. and Q. xalapensis Bonpl.), and all are capable of hosting the scale insects.

2.3.2 Bird foraging observations in oak trees Forty observational trees (five at each of two forest sites and two pasture sites) were cho- sen for study. Though it is difficult to determine specific identity accurately with vegetative characters and complications of hybridization, all individual observation trees were pre- sumed to be Quercus laurina. All sites were separated by at least 1 km. Observation trees in pasture sites and forest sites were of similar size. The average diameter at breast height (dbh) of observation trees was 52.1 cm. All observation trees harboured colonies of scale in- sects producing honeydew, although not all trees at each site had such colonies. Almost all scattered oak trees in pasture habitat contain scale insects, whereas in forest patches most oaks (>70%) bordering pasture and most oak trees in the forest interior (>50%) contain scale insects (Gamper unpubl. data). Trees with scale insects producing honeydew were easily discovered through the abundant liquid drops falling from the numerous scale insect anal filaments 3.1. The number and species of birds visiting the tree, the total time spent in the tree by each individual bird, and food items taken were recorded during 1-h observation periods at observation trees. Birds were easily determined to be feeding on scale insect honeydew. The large honeydew droplets can be viewed without difficulty using binoculars on upper branches, and without the aid of binoculars on lower branches. Scale insects colonize only the trunk and branches, therefore birds feeding upon the foliage, and using more active feeding behaviours (such as sallying) where honeydew was not present, were assumed to be feeding on insects. Since all birds could not be individually identified, repeat visits to the tree by the same individual were counted separately. Interspecific and intraspecific interactions (such as physical contact and/or chasing behaviour) were also recorded. Each individual tree was observed on four occasions (divided evenly among morning and afternoon periods) between March 2002 and May 2002, for a total of 160 h of bird observation. Observation times for each tree were randomized. Before each hour of bird observation, 10

10 a b

Figure 2.1: Morphology of scale insects producing honeydew on oak trees in trop- ical montane cloud forest of Chiconquiaco, Veracruz, Mexico. Anal filaments of these insects are visible excreting the sugary, honeydew waste. (a). Individual scale insect (Stigmacoccus garmilleri) found on oak (Quercus spp.) in late feeding instar stage. Excrement of this insect (honeydew) is visible at the end of the anal filament. (b). Colony of scale insects (Stigmacoccus garmilleri) on the trunk of oak (Quercus spp.).

11 scale insects producing honeydew on the lower trunk of the observation tree were randomly chosen, and sugar concentration and volume of the exposed honeydew drop were recorded (n = 1600). Honeydew drop volume was measured with 15-microcapillary tubes (Drummond Scientific Co.). A hand-held refractometer (Bellingham and Stanley Co., UK) was used to measure sugar concentration (and utilized in subsequent studies). Overall scale insect density at each observation tree was measured using 40 cm2 quadrat counts. Quadrat measurements (taken at 1 m and 3 m heights along the trunk of the tree and on one branch) were averaged for observation trees.

2.3.3 Statistical analysis

Mann-Whitney U-tests were used to compare the mean rank of honeydew visits for mi- gratory birds in comparison to resident species. Number of visits to observation trees and visits for honeydew consumption were compared using bivariate correlations. To examine whether the proportions of such visits were similar for all species; that is, birds visiting trees were consuming honeydew, and no individual species visited often for any other purpose; Pearson chi-square tests were used. To specify whether forests and pastures did not differ in the number of honeydew visits, the total time spent foraging on honeydew, or defen- sive chases independent-sample t-tests (equal variances were not assumed) were conducted. Pearson’s correlation coefficients were used to determine covariance between number of bird chases observed (pooled for each observation tree) and the scale insect density measures.

2.4 Results

2.4.1 Bird foraging

Fifteen resident bird species and 18 migrant species were observed visiting observation trees during the 1-h observation periods 2.1. Approximately 72% of the resident bird species and 83% of the migrant bird species observed were recorded to forage on scale- insect honeydew. The mean rank of honeydew visits was higher for migratory birds than for residents (Z = -1.97, P = 0.029, df = 42). The 21 species of birds observed to feed on honeydew at least three times made a total of 1027 honeydew foraging visits 2.1. The most frequent visitor to the honeydew was the Audubon’s warbler (Dendroica coronata auduboni; n = 272 visits), followed by the Nashville warbler (Vermivora ruficapilla; n = 214), black-throated green warbler (Dendroica virens; n = 160), and Wilson’s warbler (Wilsonia pusilla; n = 156) (Nomenclature for bird species follows Clements (2000)). Of these Nashville warblers made the highest proportion of visits during which honeydew was consumed 2.1, but each of the four fed on honeydew during more than 80% of their visits. Number of visits to observation trees and number of visits to consume honeydew were highly correlated (r2 = 0.998, P < 0.001). The proportions of such visits were similar for all species; that is, birds visiting trees were consuming honeydew, and no individual species visited often for any other purpose.

12 Table 2.1: Bird visits to scale insect honeydew on oak trees in tropical montane forest and pasture areas of Chiconquiaco, Veracruz, Mexico. The residency status (M = migrant, R = resident), number of visits to observation trees, percentage of visits during which honeydew was consumed, total time observed feeding on honeydew (HD), total time birds were observed foraging on insects, mean length (+/- SD) of honeydew foraging bouts for bird species that visited observation trees at least three times, and mean length (+/- SD) of foraging bouts for birds that were observed consuming insects. Data were accumulated from a total of 160 1-h observations of 40 observation trees.

13 2.4.2 Interspecific interactions and resource defense Honeydew was most actively defended by the Audubon’s warbler, but Nashville warbler, Wilson’s warbler and yellow-eyed junco (Junco phaeonotus) were all observed attempting to exclude conspecific birds from Quercus spp. harbouring scale insects. The bird species most commonly chased by the Audubon’s warbler was the Nashville warbler 2.2. A linear regression analysis supported the prediction of number of bird chases from scale insect density at each observation tree. As the overall scale insect density increased, the number of aggressive chases by birds increased (F1,38 = 9.28,P = 0.004). The regression equation for predicting the overall number of chases is

y = 3.48x − 3.47 (2.1)

where y = predicted chases and x = scale insect density. The correlation (r) between the index of chases and scale insect density was 0.443. Approximately 19.6% of the variation in number of aggressive chases could be accounted for by its linear relationship with scale insect density (r2).

2.4.3 Comparison of forest and pasture Twenty-three species of bird were observed to visit the observation trees located in the forest (n = 472 honeydew visits), and 34 species were observed in observation trees in the pasture areas (n = 591 honeydew visits; Table 2). Forest honeydew visits were dominated by Nashville warbler (28.6%), black-throated green warbler (22.5%), and Wilson’s warbler (21.2%). Pasture habitat honeydew visits were predominantly Audubon’s warbler (45.3%) and Nashville warbler (13.5%) (other species not listed comprised less than 9% of honeydew visits). Of 118 aggressive chases observed, only 9.65% occurred in forest observation trees, and 90.3% in pasture trees (Table 2). Audubon’s warblers were the dominant chasing bird in pasture habitats, whereas in forest habitat no bird species was as proportionately dominant in the number of total chases 2.2. Independent-samples t-tests revealed that number of bird visits for honeydew at each observation tree did not depend on habitat type. That is, forest and pasture did not differ in number of honeydew visits (t38 = -1.28, P = 0.208; 2.2) or in total time spent foraging on honeydew (t38 = −0.820,P = 0.417; Table2). They did differ significantly in number of defensive chases (t20 = −3.315,P = 0.003; equal variances were not assumed; 2.2).

2.5 Discussion

Measures of honeydew consumption and aggressive defense of the honeydew resource demonstrate the importance of scale insect honeydew as a food resource for birds. These results illustrate how aggressive defense of honeydew was much greater in pasture trees and may have been habitat driven. Greater incidence of chases observed in pasture trees was not due to facility of observation in these isolated trees. Individual oaks in sunny pasture tended to have a very dense, bushy structure while forest trees contained many open lower branches and only thick foliage in the upper branches.

14 Pasture Chases

20.6% (21)

2.9% (3) Audubon’s warbler 6.9% (7) Wilson’s warbler Chipping sparrow 1.9% (2) 48.0% (49) 1.9% (2) Ruby crowned kinglet Nashville warbler 1.9% (2) 1.9% (2) 8.8% (9) Common bush tanager Black throated green warbler 1.9% (2) Townsend’s warbler Yellow-eye junco

2.9% (3)

Forest Chases

11.1% (1) Wilson’s warbler Black-throated green warbler 33.3% (3) 11.1% (1)

11.1% (1) Nashville warbler White-eared hummingbird 11.1% (1) 11.1% (1) 11.1% (1)

Townsend’s warbler

Figure 2.2: Diagrammatic representation of interactions among birds utilizing hon- eydew on trees harbouring scale insects during 160 1-h observations on observation trees in Chiconquiaco, Veracruz, Mexico. Species observed only once have been omitted from the figure. (figure design adapted from Greenberg 1993). Arrows indicate aggressive interactions by the source species against the species to which the arrows point (interspecific chasing); curved arrows indicate intraspecific com- petition. Percentage of total chases and number of actual chases are provided. Pasture trees (above) were dominated by Audubon’s warbler, with the Nashville warbler being the most frequent object of defensive chases. Audubon’s warbler were not observed in forest trees, and the interactions among bird species were considerably different.

15 Table 2.2: Summary of birds and their feeding behavior on oak trees in tropical montane forest and pasture areas of Chiconquiaco, Veracruz, Mexico. The numbers of species, visits, honeydew (HD) visits (total number of visits to observation trees during which honeydew was consumed), time of honeydew feeding (cumulative time spent feeding on honeydew), and number of aggressive chases observed during a total of 160 1-h observation periods at 20 observation trees(equal variances not assumed) located in forest habitat and 20 trees located in pasture habitat.

trees Pasture trees t df P

Number of bird species 23 34

Number of visits 553 704 –0.1038 38 0.306

Number of HD visits 472 591 –1.282 38 0.208

Time of HD feeding (h) 56.64 69.54 –0.82 38 0.417

Number of chases 9 99 –3.315 20 0.003

16 In Chiapas, Mexico, the white-eared hummingbird (Hylocharis leucotis) is the only resident species commonly feeding on honeydew (Greenberg et al. 1993). The greater number of species found to do so in the present study ref{tab:honeydew foraging could be a result of observing both pasture and forest habitat and including a larger assemblage of bird species. Some species (bumblebee hummingbird, painted redstarts) were almost exclusively observed in forest-pasture edge habitat, one (yellow-eyed junco) in both forest-pasture edge and pasture areas, and several (Nashville warbler, Wilson’s warbler, black-throated green warbler) were mostly observed in forest habitat. The higher rate of defensive chases observed in the present study (0.712 h-1) than in a study in dry forests of the Dominican Republic (0.476 h-1; Latta et al. 2001) may result from the greater number and density of scale insects at this site than in the Dominican Republic site (Gamper pers. obs.) or other factors such as seasonality, location and bird assemblages; it is most likely due to habitat characteristics. Pasture trees alone experienced a chase rate of 1.24 h-1, in comparison with .11 h-1 for forest trees, making the forest rate for this study lower than in the dry forests of the Dominican Republic (Latta et al. 2001). Rates of defensive chases at this Veracruz site did not approach those at Chiapas, Mexico, where 12.2 chases h-1 have been recorded (Greenberg et al. 1993) and where, in addition, yellow-rumped warblers (Dendroica coronata) frequently flew from tree to tree, a behaviour recognized as territorial patrolling (Greenberg et al. 1993). We observed this common behaviour in this study, but only in pasture trees. Forest trees were much less frequently defended or patrolled. This result may be associated with the greater defensibility of isolated trees. The preferential defense of scale-insect honeydew by Audubon’s warblers in isolated pasture trees supports the hypothesis of Orians and Willson (1964) that, when forests with various structural characters are converted to more open agricultural habitats, the defensibility of resources increases. Aggressive defense may increase energy requirements and risk of injury for the individ- ual bird,. Woinarski (1984) speculated that gregarious behaviour during the non-breeding season could reduce the risk of injury for those individuals that are displaced from de- fended resources. In this system, the aggressive defense of pasture trees by Audubon’s warbler may have forced other species, such as Nashville warbler, to forage on honeydew predominantly in flocks in forested areas (Gamper, pers. obs.). In addition to increasing the chance of honeydew foraging, this flocking behaviour by Nashville warbler may reduce energy consumption and individual risk of injury. The possible resource driven changes to bird assemblages in a range of habitats compels further observation or experimentation. The occurrence of fruiting trees in fragmented landscapes can have an effect on the conservation of frugivorous bird communities (Graham et al. 2002, Levey 1988). Our results suggest that scale-infested oak trees with an abundant honeydew resource serve a similar purpose in this study site. The presence of trees with honeydew in disturbed native habitat may play a role in the maintenance of nectarivorous or omnivorous bird communities. Scattered trees in managed landscapes have been noted as keystone structures because their contribution to ecosystem functioning is disproportionately large when recognizing the small area occupied and low biomass of individual trees (Manning et al. 2006). Mature trees scattered within agricultural landscapes are important habitat for some organisms, and provide a range of ecosystem services; these trees are declining in managed agricultural landscapes globally (Gibbons et al. 2008). Since scattered trees fulfill distinctive functional

17 roles in a broad range of scattered tree ecosystems, their loss may result in unfavourable ecological regime shifts (Manning et al. 2006). Birds will be indirectly affected by habitat alterations that alter food sources, so studying bird foraging behaviour and food resource availability can aid investigators in understand- ing species vulnerabilities to habitat change (Johnson and Sherry 2001). Knowing how scale-insect colonies are altered by forest disturbance may be important for conserving the diversity of birds in this disturbed landscape. Mexico’s highland cloud-forest habitat now frequently exists only in landscapes dominated by pasture (Cayuela et al. 2006). Pasture or agricultural areas may retain a large portion of the forest biodiversity, particularly in areas with multiple land uses (Greenberg et al. 1996, Petit et al. 1995, Robbins et al. 1992). The preservation of key food resources in disturbed habitats may be important to landscape-level processes in these habitats. Birds inhabiting isolated trees in pasture areas to feed on honeydew may also offer benefits to this habitat type. Bird diversity in pasture areas may provide benefits such as pollination, pest control, and seed dispersal for growth of vegetation cover (Daily et al. 2001). Although some species depend on native forest habitat, tropical pasture areas may offer great support for maintenance of biodiversity, especially if they are managed well (Hughes et al. 2002). Our data reveal that honeydew is a very important resource for both resident and migrant birds in Veracruz, adding to previous work in Chiapas and the Dominican Republic. We found that Audubon’s Warblers exclude many other species from this resource in a pasture environment, but different dynamics occur among birds in the forest environment. Most importantly, our work demonstrates that leaving isolated oak trees in an agricultural setting can still provide an important resource for at least some species (including some Neotropical migrants). These three results are important to consider from the perspective of conservation of bird populations; preserving isolated oak trees found in pasture areas as well as conservation of the remaining tropical montane cloud forest fragments is a prescription for good management of bird species in Mexican highlands.

18 CHAPTER 3

ALTERATION OF FOREST STRUCTURE MODIFIES THE DISTRIBUTION OF SCALE INSECT, STIGMACOCCUS GARMILLERI, IN MEXICAN TROPICAL MONTANE CLOUD FORESTS

3.1 Abstract

Stigmacoccus garmilleri Foldi (Hemiptera: Margarodidae) is an ecologically important honeydew-producing scale insect associated with oak trees (Quercus spp.) in highland forests of Veracruz, Mexico. The honeydew exudates of emphS. garmilleri serve as a signif- icant nutrient source to many species of birds, insects, and sooty molds. Oak trees found in the forest interior, forest edge, and those scattered in pasture areas support scale insect colonies, though the pattern of insect infestations on trees within these varying landscape types has not been elucidated. This study aims to describe the distribution of scale insect infestation and any distinctions in honeydew production based on tree location. Scale insect density, honeydew volume, and sugar concentration were surveyed throughout a continuous landscape that included both patches of forest and scattered pasture trees. In addition, the anal filament through which the honeydew drop is secreted was also measured and was experimentally removed to test and measure regrowth. Scale insect densities on tree trunks were greatest on pasture trees, while intermediate densities were found on trees at the forest edge, and low densities on interior forest trees, suggesting that trees in disturbed areas are more susceptible to scale insect infestation. Trees with small diameters at breast height had significantly higher insect densities than trees with medium to large diameters. Trunk aspect (North, South, East, and West) was not a significant determinant of scale insect density. In forested areas higher densities of scale insects were found at three meters height in comparison to lower heights. Sugar concentrations and drop volumes of honey- dew in forest and pasture areas were not significantly different. However, scale-insect anal tubes/filaments were significantly longer in pasture than they were in forests. Sugar concen- trations of honeydew appeared to be positively correlated with temperature and negatively correlated with relative humidity. Experiments indicated that anal filaments could grow approximately 4 mm every 24 hours, and average tube growth was significantly faster in

19 pasture than in forest, suggesting that there may be a physiological effect on the insect due to landscape disturbance. The results obtained in this study describe the increases in scale insect infestation of trees with forest disturbance. The effect of these increased scale insect densities on the host tree physiology is still to be resolved.

3.2 Introduction

The montane cloud forests of northeastern Mexico have a high concentration of en- demism and are increasingly vulnerable to climate change, deforestation, and habitat frag- mentation (Markham 1998; Bubb et al. 2004). Ninety percent of the original montane cloud forest in Mexico has been lost (Ramirez-Marcial et al. 2001). Remaining forests are discontinuously distributed and commonly form an abrupt boundary with adjacent grazed grassland, small agricultural fields, and clearings for houses and community structures. As habitat fragmentation becomes more pervasive throughout the world, our understanding of forest fragmentation has also grown more sensitive to context. The species interactions initiated with fragmentation and post-fragmentation disturbance regimes may magnify the impacts of fragmentation (Laurance 2002; Ewers and Didham 2006; Malanson et al. 2007). A notable feature of the highly fragmented montane forests of central Veracruz, Mex- ico, is the interaction between oak trees (Quercus spp.) and phloem-feeding scale insects (Gamper and Koptur 2010), identified as Stigmacoccus garmilleri Foldi (Hemiptera: Mar- garodidae) (Foldi 1995; Hodgson et al. 2007). Immature S. garmilleri instars colonize trunks and branches by burrowing under tree bark. The scale insects insert their mouth- parts, called stylets, into phloem cells and feed on phloem. The phloem of the host plant is rich in carbohydrates but low in compounds containing soluble nitrogen and amino acids, which are necessary to the insects for protein building (Gullan and Kosztarab 1997). Phloem feeding insects therefore ingest and excrete large quantities of carbohydrates in the process of acquiring sufficient amino acids (Wckers 2000). This waste excretion, termed honeydew, forms droplets at the end of long anal tubes, or anal filaments 3.1. Honeydew-producing insects tend to eliminate copious honeydew, live in groups, and are typically sedentary or semi-sedentary (Williams and Williams 1980). For several months of the year S. garmilleri resides within the tree in the form of these feeding instars. Adult female and male insects develop and can be found mating on the surface of the tree 3.2. Details of the life cycle of S. garmilleri were described by Hodgson et al (2007). In some ecosystems, honeydew forms important ecological links within trophic levels and occasionally represent the primary carbohydrate food source for diverse group of organisms (Grant and Beggs 1989). Birds are commonly found foraging on and defending the rich and abundant honeydew resource originating from S. garmilleri (Gamper and Koptur 2010). The honeydew of S. garmilleri also provides food for a varied community of arthropods such as ants, vespid wasps 3.3, honeybees (Apis mellifera), mites (Anystis sp.) (Figure 3B), and dipteran species. Furthermore, honeydew provides nourishment for a dense growth of black sooty mold that in turn provides habitat for many invertebrates (Didham 1993) and may provide nourishment for bacteria and fungi that decompose the litter on the forest floor (Morales 1991), although Wardhaugh and Didham (2006) found lower decomposition rates in high honeydew areas.

20 Figure 3.1: Hair-like anal filaments from scale insect Stigmacoccus garmilleri and drops of honeydew secreted from their ends (Chiconquiaco, Veracruz, Mexico). Scale insects are capable of reaching high densities on oak tree trunks and branches.

21 Figure 3.2: Winged adult male (Stigmacoccus garmilleri) mating with soft-bodied adult female. For approximately three weeks during their life cycle adults are found mating on the trunks of host trees (Quercus spp.) in tropical montane forests of central Veracruz, Mexico.

22 Figure 3.3: Invertebrate species are known to feed on honeydew produced by scale insect Stigmacoccus garmilleri in tropical montane forests of central Veracruz, Mexico. (a) Mite (Anystis sp.) found foraging on a honeydew droplet and (b) Vespid wasp in flight consuming honeydew.

23 Despite the importance of honeydew in montane forests of Mexico, little is known about the distribution of scale insect colonies in the present landscape. Insects that are concealed while feeding, such as sap-sucking scale insects, stem-borers, and leaf miners, are likely to have heterogeneous spatial distributions in the host environment (Wardhaugh et al. 2006). Changes to the forest structure may alter the density and distribution of scale insects and the honeydew resource. Local factors such as tree species, tree size, growth rate, genotype, and exposure are believed to generate the large population variances observed between adjacent trees (Gaze and Clout 1983; Kelly 1990; Didham 1993). Individual trees can also be vertically stratified into discrete zones or layers from the roots to the crown (Didham and Fagan 2004). In New Zealand beech forests the within-tree distribution of the sooty beech scale was vertically stratified with the greatest insect densities occurring on bark surfaces in the canopy instead of on lower trunk surfaces (Wardhaugh et al. 2006). The study presented here investigates populations of S. garmilleri, a scale insect whose stable persistence benefits an array of species in a Mexican tropical montane cloud forest habitat that has undergone considerable land-use change. The main objective of this work is to compare the distribution of scale insects and their honeydew production on the host tree in three distinct habitats, namely forest interior, forest edge, and isolated pasture.

3.3 Methods 3.3.1 Study site The study was conducted in tropical montane forest near the town of Chiconquiaco, in the state of Veracruz, Mexico 3.4. The area has three seasons: moderately dry and cool from October to March, dry and warm from April to May, and wet and warm from June to September. Mean annual temperature is 15.2 C and total mean annual precipitation is 1532 mm) (Williams-Linera et al. 2000). The area was covered by heavy fog on most days, and humidity levels typically remain high even during the relatively dry months. The elevation was approximately 2000 m, and the habitat was a mosaic of mature forest patches, cattle pastures, and small cornfields. The study trees were located in two 25 ha forest patches and two 35 ha pasture areas all on west-facing slopes. All sites were separated by at least 1 km. Forest areas had closed canopies dominated by Quercus spp., and pasture areas were open and included a few, large, scattered Quercus spp. individuals. Forest fragments had low edge to interior area ratios and were comprised of a canopy of hybridizing assemblages of Q. laurina Bonpl., Q. germana Schltdl. Cham., Q. salicifolia Ne, Q. corrugata Hook., emph{Q. affinis Schweid., and Q. xalapensis Bonpl., all of which are capable of hosting the scale insect S. garmilleri.

3.3.2 Scale insect density measures For determination of scale-insect densities, insects were counted on oak trees in nine randomly chosen 10 x 10 m plots; three plots each in forest interior, forest edge, and pasture. Though it is difficult to determine species identity accurately with vegetative characters and complications of hybridization, all oak trees on which scale insect density was recorded were presumed to be Quercus laurina. In these plots, the diameters at breast height of all Q. laurina trees over 2 m tall were recorded. Anal tube density was determined as number of

24 Figure 3.4: The study was conducted in tropical montane forest near the town of Chiconquiaco, in the state of Veracruz, Mexico

25 filaments occurring within copper wire frames; 20 x 20 cm for trunks and large branches, 10 x 40 cm for thinner branches. Insect counts within these wire frames were conducted at 1m, 2m and 3m on the north, south, east, and west sides of the tree for a total of 12 measurements per tree. Other studies in New Zealand have shown that 90% of filaments are connected to feeding larvae (Moller and Tilley 1989), so anal-tube presence was taken as a proxy for insect presence. It is not specifically known whether the presence of an anal filament indicates that S. garmilleri is actually alive. Anal filaments producing honeydew were assumed to be attached to living insects. Filaments without drops may still be connected to a living insect, albeit not producing honeydew at the time. The number of filaments with honeydew drops within each frame was also recorded. To determine the overall insect density, we used the sum of the 12 insect counts per each 400 cm2 sample frame for a total of 4800 cm2 sampled. On the basis of diameters at breast height measurements, these trees were categorized as small (5-28.6 cm), medium (29-46.3 cm), or large (48.9-83 cm); each category included roughly equal numbers of trees when averaged over the three habitats. Pasture areas only included small and medium sized trees, whereas forest and forest edge habitats included oaks in all size classes. Insect density data were not normally distributed, but population distributions were uniform among samples. Kruskal-Wallis nonparametric comparison tests were used to determine whether tree height, aspect, and tree-diameter size class affected density of scale insects. Multiple comparison tests were conducted on pair wise differences among groups. Type I error across tests was controlled by the Holm’s sequential Bonferroni approach.

3.3.3 Anal filament and honeydew drop measures at observation trees

Forty observational trees were chosen for the study; ten at each of two forest sites and two pasture sites. All sites were separated by at least 1 km. All observation trees harbored colonies of scale insects producing honeydew, although not every tree at each site had such colonies. Each individual tree was observed on four occasions, divided evenly among morn- ing and afternoon periods between March 2002 and May 2002. Observation times for each tree were randomized. At each tree, 10 scale insects producing honeydew on the lower trunk of the observation tree were randomly chosen, and sugar concentration and volume of the exposed honeydew drop were recorded in addition to the anal filament length (n = 1600). Honeydew drop volume was measured with 15µl microcapillary tubes (Drummond Scien- tific Company, www.drummondsci.com). A hand-held refractometer (Bellingham Stanley, www.bellinghamandstanley.com) designed for volumes as small as 0.5 µl was used to mea- sure sugar concentration in honeydew. The presence of amino acids in honeydew may contribute to the refractive index, but this effect was considered negligible (Inouye et al. 1980). Refractivity is a reliable measure of the sugar concentration of honeydew drops (Grant and Beggs 1989). Only honeydew from ground level to 3 m was sampled. Ambient air temperature and relative humidity were also recorded when measurements were made. Lengths of scale insect anal filaments were measured using a digital caliper. Independent-samples t-tests were used to test for differences in anal filament length by habitat type. Correlation analysis was used to examine the relationships among sugar concentration, honeydew volume, relative humidity, temperature, and anal-tube length.

26 3.3.4 Anal filament growth rate at observation trees Half of the original 40 observation trees were randomly chosen as experimental trees for examination of anal tube growth rate. At each tree 12 scale insects were individually marked with colored pins just above their position on the tree. The anal filaments originating from scale insects were measured using digital calipers and were then experimentally removed. Only insects producing honeydew drops at initiation of the experiment were chosen for study to ensure that living insects were marked. The following day anal filaments were re-measured and average growth over a 24 hour time period was calculated. Correlations between the original length of each anal tube and its rate of re-growth were calculated. Independent-samples t-tests were used to test for differences in tube growth between pasture and forest. Some originally marked insects were not recovered during the re-measurement period due to pin drop from oak trunks.

3.4 Results Scale insect density measures The overall mean number of anal filaments per 400 cm2 frame was 13.4 (±20.5SD). The mean percentage of productive scale insects (filaments with honeydew drops/total number of anal filaments) at the time of sampling per 400 cm2 was 36.4% (±34.2SD). Because the presence of a honeydew drop may be influenced by wind or rain, this figure may be an underestimate. We therefore used anal-tube densities as the measure of scale insect abundance. No significant difference in anal tube density among the four aspects (north, south, east, and west) in any habitat (p > 0.05 in all cases; Kruskal-Wallis test) were found. The density of insects among the three trunk heights was not significantly different in edge habitat (χ2 = 0.70, p = 0.71) or in the pasture area (χ2 = 0.001, p = 0.99), although there was a significant difference in forest habitat (χ2 = 10.98, emphp < 0.01). Trees in the forest had greater densities of scale insects at three meters height in comparison to lower heights 3.5. Anal-tube density was highest in the pasture area (x = 16.5, 30.9 SD), followed by edge trees (x = 14.8, 18.4 SD) and forest trees ( x = 10.5, 14.9 SD). These differences were not significant (p > 0.05 in all cases; Kruskal-Wallis test). The three tree diameter classes differed significantly in median scale-insect densities (χ2 = 7.91, p < 0.05). Smaller trees harbored greater densities of scale insects than did medium and large trees ??. Multiple comparison tests revealed the small and medium diameters at breast height classes differed significantly (p < 0.05), as did the small and large classes (p < 0.05). The medium size class did not differ significantly from the large class (p = 0.831).

3.4.1 Anal filament and honeydew drop measurements On average, scale insects associated with forest trees had longer anal filaments than those on forest edges or in pastures. Independent-samples t-tests revealed that habitats differed significantly in scale insect anal-tube length; mean tube length was on average 5.2 mm greater in forest than in pasture areas 3.1. Habitats did not differ significantly in sugar concentrations or in the volume of honeydew 3.1. Correlation analysis was used to examine the relationships among sugar concentration, honeydew volume, relative humidity, temperature, and anal-tube length ( 3.2; 3.7; 3.8).

27 Figure 3.5: Mean estimates (95% confidence interval, standard deviation) of scale- insect density per 400 cm2 on trees at different heights in forest habitat.

28 Figure 3.6: Mean estimates (95% confidence interval, standard deviation) of scale- insect density per 400 cm2 on trees of different size classes within forested habitat. Small = 5-28.6 cm, medium = 29-46.3 cm, large = 48.9-83 cm diameters at breast height

29 Table 3.1: Means, standard deviations, minimum and maximum values for honeydew-drop volume, sugar concentrations, and anal-tube lengths for scale in- sects (Stigmacoccus garmilleri) on oaks in forest (n = 756) and pasture habitats (n = 762) of Chiconquiaco, Mexico. Independent-samples t-tests revealed that habitats differed significantly in anal-tube length; however habitats did not differ significantly in sugar concentrations or volume of honeydew. *Significant at the 0.01 level

Variables Habitat N Mean Min Max SD t p Volume (ul) Forest 756 0.64 0.11 4.5 0.53 0.495 0.387 Pasture 762 0.65 0.11 11.11 0.66

Sugar (%) Forest 756 33.07 3 80 17.67 1.278 0.387 Pasture 762 34.24 2 100 18.02

Tube Length (mm) Forest 756 32.07 2 100 17.77 5.862 * 0.003 Pasture 762 27.07 1 127 16.75

30 Table 3.2: Spearman’s correlation coefficients were calculated for the predictive strength of the recorded variables (temperature, relative humidity, drop volume, sugar concentration, and anal-tube length). Data were combined from 20 forest trees and 20 trees located in pasture habitat in Chiconquiaco, Mexico. In all, 1518 insects producing honeydew were examined, and data on temperature and humidity were recorded when sugar concentration, drop volume, and tube length were measured. *Significant at the 0.005 level.

Temperature Humidity Volume % Sugar

Humidity –0.659*

Volume –0.045 –0.107*

% Sugar 0.418* –0.702* 0.219*

Tube length –0.047 0.003 0.122* 0.035

The Bonferroni approach was used to control for Type I error across the 10 correlations; a p value of less than 0.005 was required for significance. The results show that 6 out of 10 correlations were statistically significant. Honeydew-drop sugar concentrations were posi- tively correlated with increases in temperature 3.7 and negatively correlated with relative humidity 3.8.

3.4.2 Anal filament growth Prior to removal of anal-tube filaments, scale insects used for measurements of anal- tube regrowth had a mean anal-tube length of 33.56 mm, with a large standard deviation of 20.84 mm. Twenty-four hours after anal filaments were removed the new filaments had reached a mean length of 3.89 mm ( ±1.59SD). An independent-sample t-test, not assuming equal variances, revealed that that forest and pasture trees differed in both original mean tube length and mean tube length 24 hours after tube removal. Insects in forest trees had significantly longer original anal filaments than did those in pasture trees (t115 = −3.81,

31 Figure 3.7: Honeydew sugar concentration plotted against temperature recorded at time of measurement. Data were collected from 20 trees in forest habitat and 20 trees located in pasture habitat in Chiconquiaco, Mexico. n = 1518 insects producing honeydew were recorded between 15 March 2002 and 17 April 2002.

32 Figure 3.8: Honeydew sugar concentration plotted against relative humidity recorded at time of measurement. Data were collected from 20 trees in forest habitat and 20 trees located in pasture habitat in Chiconquiaco, Mexico. n = 1518 insects producing honeydew were recorded between 15 March 2002 and 17 April 2002.

33 p < 0.01), and those in pasture trees had significantly higher rates of anal-tube regrowth (t193 = −2.09,p < 0.05). The correlation between rate of regrowth and original length was low (Pearson correlation coefficient; r = 0.056) and not statistically significant (p = 0.416).

3.5 Discussion

Data reported here reveal differences in the biology of scale insects depending on habitat type. In Chiconquiaco and in many other areas of Mexico, highland tropical forest habitat exists in a landscape dominated by pasture. The maintenance of key food resources like scale-insect honeydew in these disturbed habitats can be important to maintaining the animal community and critical landscape-level processes.

3.5.1 Scale insect distribution Knowledge of the spatial distribution and abundance of S. garmilleri within host trees is essential to understanding how land-use change can affect biotic insect-plant interactions, which in turn can modify forest dynamics. Populations of S. garmilleri are ideal for testing within-tree heterogeneity in herbivore densities because feeding instars are embedded and sedentary within the bark of the tree, and their spatial distribution is not a transitory phenomenon as it would be in the evaluation of free-living herbivores. There is strong evidence of a stationary spatial distribution in beech scale insects, which share similar natural history characters with S. garmilleri, over time periods of several months to one year (Kelly et al. 1992; Murphy and Kelly 2003). Variables that might affect the distribution of scale insects within single trees include al- titude, host species, exposure to sunlight, aspect (north, south, east, west), trunk diameter, and possibly region (Kelly 1990; Dungan and Kelly 2003). Crozier (1981) found higher den- sities of scale insects on northern aspects of Nothofagus solandri trunks and hypothesized that higher mean temperatures on this side contributed to the effect. In agreement with Kelly (1990), aspect was not correlated with anal-tube density in our study. The varying slopes and corresponding microclimate created on tree trunks, irrespective of aspect, may have confounded the findings in this study. Scale insect densities have been reported to be higher on edge-habitat trees in Nothofagus forests of New Zealand, although these differences were not quantified (Crozier 1981). In the present study, edge trees had greater densities of scale insects than forest areas, and densities were highest in pasture areas. Wardhaugh et al. (2006) recognized the greatest densities occurring occurred on bark surfaces in the canopy rather than on the trunk and on the lower rather than upper sides of the branches. Our findings suggest that trees in pasture areas and on the forest edge have dense colonies at all heights measured, whereas scale insect colonies in the forest interior had significantly greater numbers of scale insects at the highest height measured. Tree size and scale insect densities were inversely related with small trees harboring the highest densities, supporting Wardhaugh (2006) who noted a decrease in scale insect density with increasing diameters at breast height or branch diameter. In contrast, Kelly (1990) found that intermediate-sized trees harbored significantly greater tube densities while the small trees and thin upper branches bore almost no scale insects. Even the thinnest branches

34 of oak trees in Chiconquiaco were found to frequently harbor scale insects. Physiological differences between the insect species and/or tree species in the distinct geographic areas of New Zealand and Mexico may cause this difference. Wardhaugh et al. (2006) discuss how insect density differences in relation to tree size may be driven by a positive correlation between bark thickness and diameter. A number of studies have found that the establishment of sap-sucking insects was influenced by bark characteristics (Geraud-Pouey et al. 2001). Oak trees that supported no scale insects on the trunk likely had a low availability of establishment sites. It is possible that newly created fissures were inhabited by scale insects because they were swiftly colonized by crawlers, the dispersal stage of S. garmilleri. Processes that result in cracking of the bark surface such as increased exposure to the sun (Nicolai 1986) or rapid/released growth are both processes that are more predominant after deforestation of nearby trees. Forest loss and land use change may increase the number of available establishment sites within remaining trees.

3.5.2 Honeydew volume/concentration The highest honeydew sugar concentrations were found when humidity was low and temperature high. In warm sun, droplets become more concentrated as water evaporates. In forests of New Zealand, morning sugar concentrations were lower because less water had evaporated from the honeydew. Honeydew was more concentrated on hot days with dry winds (Dungan and Kelly 2003). This was not always the case in our study area; afternoons were often characterized by an increase in fog and mist and a decrease in temperature. A dynamic model by James et al. (2007) of honeydew droplet production by sooty-beech scale insects predicts different types of behavior depending on local environmental condi- tions. Honeydew droplets are highly concentrated in dry conditions with high evaporation rates, complicating further excretion resulting in cessation of droplet formation. In humid conditions, droplet formation continues indefinitely (James et al. 2007). Dungan et al. (2007) found that rates of production were significantly related to en- vironmental conditions over the hours preceding measurement, with air temperature and air saturation deficit averaged over the preceding 24 and 12 hours, respectively. The con- tribution of temperature and humidity to variability in honeydew production illustrates a strong influence of environmental conditions. On average, pasture areas were warmer and less humid than forested areas. The lack of significant differences in this study among habi- tats in honeydew sugar concentration and volume does not rule out effects by changes in land management practices on honeydew production by scale insects. Due to the temporal variability of these environmental conditions, monitoring insects and honeydew production simultaneously in the varying habitats is suggested for future studies.

3.5.3 Anal filaments The finding of significantly longer anal filaments in forest trees compared to pasture trees may have been due to greater exposure of the pasture trees to winds and rain, which can break filaments. Filaments also may be broken by wasps (Moller and Tilley 1989) or birds (Greenberg et al. 1993; Latta et al. 2001) feeding on the honeydew. Filament breakage inside protective netting placed around S. garmilleri colonies in Chiapas, Mexico, was only 30%,compared to 70% in an unprotected sample during the same period (Greenberg et al.

35 1993). In Chiconquiaco birds occasionally broke filaments while feeding on honeydew, and cows were observed rubbing against pasture trees, contributing to breakage low on the tree trunks in pasture areas. In the present study, re-growth of broken filaments on scale insects occurred at an average rate of 4 mm per 24 hours, a previously undocumented phenomenon. A tube of average length can therefore be replaced within about seven days, assuming a constant growth rate. Breakage is apparently not detrimental to the insects; they can produce a tube long enough to secrete honeydew drops within 24 hours. The higher rate of tube growth in pasture trees than in forest trees may be attributable to higher temperatures and greater sunlight availability.

3.6 Conclusions

Scale insect honeydew is an important food source for many organisms inhabiting mon- tane cloud forests (Gamper and Koptur 2010). Honeydew is particularly essential for nectar feeding organisms when nectar resources are scarce (Murphy and Kelly 2003). Changes in availability and distribution of honeydew could have profound effects on a large community of other organisms. Examination of the distribution of scale insects in this system indicates that mosaics of forest and pasture provide good habitat for S. garmilleri. The great abundance of honey- dew on scattered pasture trees may be of particular importance in this region. Scattered pasture trees have been found to be keystone structures because of the disproportionally large ecosystem services they provide relative to the area they occupy, in addition to the maintenance of habitat and connectivity to other habitat types (Manning et al. 2009). Scat- tered trees are threatened in many places making appropriate levels of tree regeneration and preclusion of premature mortality of mature trees essential for maintaining these trees within landscapes (Gibbons et al. 2008). Despite the benefits of habitat and food resources to other organisms, increases of scale insects on scattered pasture and forest edge trees may induce physiological stress, transform forest growth dynamics, and decrease reforestation potential (Chapter 4). The spatial distribution of scale insect populations on trunks and branches of trees of increasing diameters at breast height may indicate a strong temporal component to the spatial dynamics of scale insects driven by changing host tree phenology. Future studies on phytophagous insects infesting large host trees should consider more explicitly changes in population dynamics both spatially and temporally (Wardhaugh et al. 2006). In or- der to detect interactions between tree size and location (interior, edge, pasture), future experimental design should incorporate information on geographic location, with accompa- nying scale insect density and other descriptive attributes for a thorough, spatially-explicit statistical analysis. Understanding more about the dispersal of scale insects and potential genetic predispo- sition for some individual trees to be more susceptible to scale insect populations may also be fundamental in understanding how they will be distributed in a changing landscape, and how the modifications of the landscape will alter their distribution. In cases of dispersal limitation, isolated pasture trees could lead to deleterious changes in the insect popula-

36 tion genetic structure. These potential directions for future research may yield insight into unanswered questions about this important food resource in tropical montane cloud forests.

37 CHAPTER 4

EXPLAINING THE RELATIONSHIP BETWEEN CHRONIC SCALE INSECT HERBIVORY AND TREE GROWTH IN FRAGMENTED FORESTS OF VERACRUZ, MEXICO

4.1 Abstract

The herbivorous scale insect, Stigmacoccus garmilleri, can attain high densities in oak trees in tropical montane forests of Mexico. These insects feed on phloem tissue for pro- longed periods of time and can infest trees across the extent of forest patches and entire landscapes (estimated square kilometers). To date, the effects of long-term removal of vas- cular tissue and chronic herbivory on plants remain understudied. A pattern of decreasing tree growth rates with increasing scale densities should be expected when considering po- tential costs of phloem removal to oak trees. However, this potential cost should be offset by growing conditions that vary not only according to the density of scale insects, but also according to tree spatial location within remaining forest fragments. Using dendrochrono- logical analysis, tree ring-based growth rates were developed for trees with varying levels of scale insect infestations and with distributions in the forest interior and along the edge of a forest patch. Large trees in the sample exhibited larger average annual growth than smaller trees. Both large and small trees without heavy infestations of scale insects grew better than similar size classes that had minimal infestations of scale insects. Trees growing on the forest edge did not have any appreciable growth differences in relation to level of scale insect infestation. This lack of relationship may be due to the fact that edge trees are growing under more sunlit conditions minimizing any observable differences in average growth due to impacts of scale insects. Scarring to tree tissue was evident in the tree cores by the appearance of ’scale flecks’. The presence of scale flecks did not correlate with variations in tree growth, however the appearance of these flecks in spatial location was clustered and merits further investigation.

38 4.2 Introduction

Insects can be influential in determining the overall physiological health, form and per- formance of woody plants. Plant feeding or herbivorous insects can alter the species com- position, ecosystem function, and socioeconomic value of forests. For example, mountain pine beetles (Dendroctonus sp.) in the western U.S. have altered fire regimes and forest management practices (Kashian et al. 2011, Kayes and Tinker 2011). Infestation by the hemlock woolly adelgid (Adelges tsugae) in forests of the Appalachians led to landscape- level changes in overstory and understory forest composition (Spaulding and Rieske 2010). However, not all herbivorous insects have such acute effects, and their action is more chronic over a longer period of time. Although herbivores are integral part of forests, under some conditions they can produce undesirable effects and degradation of forest resources (Ayres and Lombardero 2000). Very few studies have investigated the effects of a stressor such as chronic herbivory on long lived plants (Yang and Karban 2009). Of the existing studies, several have shown sub- stantial declines in host growth and performance associated with chronic herbivory. Both Morrow and LaMarche (1978) and Trotter et al. (2002) illustrated that chronic aboveground insect herbivory significantly reduced the growth of trees in forested systems. A transition matrix developed by Doak (1992) suggests at even low intensities, chronic herbivory can substantially reduce the growth of long-lived plants. The effects of stress are not necessarily undesirable (e.g. may sometimes include an increase in tree defenses (Karban and Bald- win 1997) or an increase in forest productivity (Teskey 1997). Some examples of chronic herbivory by phloem feeding insects suggest that while the insects harvest relatively large amounts of carbohydrate from their host trees, the consequences of this for tree growth and reproduction may be small (Dungan et al. 2007) The majority of these studies and most investigations of effects of herbivory on tree growth in general have relied upon a variety of techniques including photosynthetic modeling, gas exchange studies, and mea- sures of induced resistance. Far fewer studies examine the potential impacts of herbivores using tree-ring analysis or dendrochronology. We adopted a dendrochronological approach to studying long-term impacts of chronic above ground herbivory by scale insects on oak trees. Dendrochronology is particularly useful in lending an understanding to processes on longer time scales. Several studies have used dendrochronology to focus on root parasites such as periodical cicadas that feed on xylem (Koenig and Liebhold 2003, Yang and Karban 2009, Speer et al. 2010). The periodical cicada system is somewhat analogous to the scale insect - oak tree system in this study particularly because trees undergo herbivorous tissue removal for many months at a time. No studies to date have used tree rings to understand the effects of chronic aboveground phloem removal on trees. Most dendroentomological research (using tree rings to date and study the past dynamics of insect populations) has focused on the loss of photosynthetic tissue in the forms of buds and leaves and potential corresponding changes to tree growth. (e.g. studies of: spruce bud worm, tussock moth, and Pandora moth outbreaks)(Speer et al. 2010). Investigations of periodical cicadas inhabiting forest areas suggest that cicada herbivory can slow the growth of host trees (Karban 1980, Koenig and Liebhold 2003). The understood primary relationship between these observations is that cicadas and their host trees have an

39 antagonistic host-parasite interaction leading to an expectation of a negative relationship between nymphal cicada densities and the growth of their host trees. Both Yang and Karban (2009) and Speer et al. (2010) investigated the effects of below ground, chronic, long- term herbivory of periodical cicadas (Magicicada spp.). The effects of cicada herbivory were most apparent at relatively high cicada densities (Yang and Karban 2009) and no effect from root parasitism was evident in any of the tree species studied by Speer et al. (2010) although there was a significant reduction in growth found the year of or the year after the emergence year of cicadas. The goal of this study is to examine the feasibility of using tree rings to characterize the effects of long-term herbivory in highly fragmented montane forests of central Veracruz, Mexico. These forests are host to a widely-observed interaction between oak trees (Quercus spp.) and phloem-feeding herbivorous scale insects (Gamper and Koptur 2010), identified as Stigmacoccus garmilleri Foldi (Hemiptera: Margarodidae) (Hodgson et al. 2007). Immature S. garmilleri instars colonize trunks and branches by burrowing under tree bark. The scale insects insert their mouthparts, called stylets, into phloem cells and feed on phloem. The phloem of the host plant is rich in carbohydrates but low in compounds containing soluble nitrogen and amino acids, which are necessary to the insects for protein building (Gullan and Kosztarab 1997). Phloem feeding insects therefore ingest and excrete large quantities of carbohydrates in the process of acquiring sufficient amino acids (Wackers 2000). This waste excretion, termed honeydew, forms droplets at the end of long anal tubes, or anal filaments. Honeydew-producing insects tend to eliminate copious honeydew, live in groups, and are typically sedentary or semi-sedentary (Williams and Williams 1980). For several months of the year S. garmilleri resides within the tree in the form of these feeding instars. Adult female and male insects develop and can be found mating on the surface of the tree for several weeks time. After mating, females lay eggs in the cracks and creviced of oak trees. These eggs hatch to the crawling or crawler stage of S. garmilleri in which the insect makes its way to deep crevices to settle and feed. Details of the life cycle of S. garmilleri are further described by Hodgson et al. (2007). The specific questions we investigate pertain to the relationships among insect density, light levels, and tree growth. We are interested in understanding how the forest fragmen- tation process changes the relationship between scale insects and host tree. How do young trees located on the forest edge with increased sunlight and increased densities of scale in- sects grow in comparison to smaller trees growing in the forest understory? Do older trees grow differently with dense infestations of scale insects? These are a few of the questions we attempt to answer.

4.2.1 Forest change and scale insect distribution Specimens collected of S. garmilleri have been collected from southern Chiapas state (Greenberg et al. 1997) and eastern Veracruz state (Hodgson et al. 2007). Reports of similar insects likely to be S. garmilleri have been reported from the northern most section of tropical montane forest in Tamaulipas state (Edwards 1982). The distribution of S. garmilleri may span most of continental Mexico within tropical montane oak forest. When tropical montane forest was more continuous and less fragmented, insects were likely limited to upper branches of mature trees with available sunlight. In current remaining pieces of

40 forest, insects in forest interior were found to be densest in upper branches (Gamper et al. 2011; chapter 3). The limited distribution on forest trees in intact forest may create a scenario where photosynthetic losses to tree are negligible to tree growth. With the creation of extensive forest edge through the process of forest fragmentation, insect densities on edge trees are now significantly greater (Gamper et al. 2011; chapter 3) and preferable, inhabitable tree surface now includes lower trunk up to upper branches in sunlit edge trees. Similar microscale distributions have been observed for cicadas. The high abundance of forest edges, which are highly preferred by periodical cicadas (Karban 1980) can produce high concentration of cicadas within relatively small areas of forested land and increase par- asite pressure on fragmented and urban forest communities (Medley et al. 2003, Yang 2006). Young trees in Chiconquiaco regenerating on forest edge may face greater colonization rates by scale insects. This may be partly determined by the oviposition decisions of the adult females (preference for sunlight) and partly because pest populations commonly increase with increasing disturbance because disturbance tends to favor fast-growing plant species, which tend to be poorly defended against herbivores (Coley et al. 1985). How these young developing trees handle photosynthetic loss in comparison to larger more established forest edge trees with high densities of scale insect poses much interest for study especially within the light of reforestation and habitat restoration. We attempt to address this question by including small edge trees with a range of levels of scale insect infestation into our tree core sampling. Herbivory and the selection of habitat are both important processes that affect the dy- namics between scale insects and their oak host trees (Fig:Dendrogrowth) These processes however suggest alternative hypotheses for the relationship between scale insect density and host tree growth in montane oak forests of Mexico. If this interaction is largely influenced by damage of chronic herbivory then oak tree growth should decline with increasing scale insect density (H1). If this association is largely influenced by the habitat selection deci- sions of ovipositing female scale insects, oak tree growth and scale insect density should be positively correlated due to the positive effects of high-light environments of both scale insects oviposition density and tree growth (H2). The observation of no significant correla- tion between scale insect density and tree growth would support the null hypothesis (H0). Here we report the findings of field studies focused on evaluating these hypotheses, using tree increment cores, scale insect density values, and mapped tree locations.

4.3 Methods 4.3.1 Dendrochronology in the tropics The presence of annual growth rings in tropical trees has been denied for a long time (Lieberman et al. 1985) This idea was mainly attributed as a consequence of a lack of clear seasonality in the humid tropics (Rozendaal and Zuidema 2011); however the existence of annual growth rings in tropical trees was described as early as 1927 (Coster 1927). It is know well understood that tropical trees undergo cambial dormancy and form annual rings due to unfavorable growing conditions. Dry season (Worbes 1999) flooding in floodplain forest (Schngart et al. 2002), salinity fluctuations in mangrove forests (Chowdhury et al. 2008), and seasonal precipitation changes (Lopez and Villalba 2011), are all periods in which annual

41 Rate of tree growth

(-) herbivory removal of (+) photosynthesis photosynthetic material Scale insect Light environment density (+) habitat selection

Figure 4.1: How tree growth rates may be influenced by the light environment and scale insect density in tropical montane cloud forests of Chiconquiaco, Mexico. Arrow direction indicates the pathway to this influence. Figure adapted from Yang and Karban (2009).

rings are formed in tropical trees. Thus part of this study is the establishment of whether or not the oaks in these montane forests have a ring signature amenable for constructing annual growth.

Tropical tree species have been identified as useful for developing long chronologies to aid in climate reconstruction (Stahle 1999). Tropical tree species can also providing stable isotope measurements to understand response of tropical trees to climate variation and change. These studies can contribute to our knowledge of role of tropical forests in the global carbon cycle, forest canopy growth trajectories, quantification of autocorrelated tree growth, and contributed to estimates of tree ages in tropical forests (Rozendaal and Zuidema 2011). Tropical dendroecological studies that are lacking involve investigations of the causes and consequences of growth variation within and among trees and their relation to environmental variation (Rozendaal and Zuidema 2011). The research reported here contributes to our improved understanding of what factors in tropical forests can contribute to growth variation.

42 4.3.2 Study area The study area is located 5 km east of Chiconquiaco, Veracruz, Mexico at approximately 2000 m. This site is characterized by three seasons: moderately dry and cool from October to March, dry and warm from April to May, and wet and warm from June to September. Mean annual temperature is 15.2 C and total mean annual precipitation is 1532 mm) (Williams- Linera et al. 2000)( 4.2). The area was covered by heavy fog on most days, and humidity levels typically remain high even during the relatively dry months.

Figure 4.2: Climograph illustrating average monthly precipitation and tempera- ture (1996-2003) in Naolinco de Victoria, Veracruz (10 km from study site). Cli- mate data are from http://www.csva.gob.mx

4.3.3 Field data collection Tree core samples were collected in July of 2008 to examine the relationship between scale insect densities and long-term host tree growth rates. Two oak stands were selected that contained oak trees with a range of scale insect densities. One stand (BB) was located on a southern slope in a forest area with light to moderate goat grazing (19.756607, - 96.786675) and with moderate natural regeneration. The other stand (PP) was also located on a southern slope in an area that has frequent grazing by cows in the understory and is dominated by widely spaced large oak trees with no or very limited natural regeneration (19.752164, -96.778221) The horizontal distance between the two sites is approximately 1

43 km. Forest area PP can be characterized by a dominance of older oak trees that had been considerably thinned through the removal of large trees (as evident in existing tree stumps), heavily grazed understory, with little to no new tree establishment in the understory or forest edge 4.3. Forest area BB has more diverse range of age and size classes, less grazing present in the understory and younger oaks regenerating on the forest edge 4.3.

A B

Figure 4.3: The two forest areas from which tree core samples were collected:(A) Forest area PP showing a heavily grazed understory, with little to no new tree establishment in the understory or forest edge, (B) Forest area BB showing a more diverse range of age and size classes and less grazing present in the understory.

All overstory oaks (Quercus laurina, >5 cm DBH) within each circular 20 m radius plot were cored at Diameter Breast Height (DBH) with an increment borer. Two plots adjacent to each other were sampled in each area (PP and BB) for a total of four forest plots. We followed methodology described in Fritts (2001) and Stokes and Smiley (1968) for the collection and preparation of samples and subsequent tree-ring chronology development. Two cores from each tree were extracted at approximately 1.3 meters on opposite sides using an increment borer. The following numbers of oak trees were cored within the four individual plots: 15 (PP Plot 1), 10 (PP Plot 2), 5 (BB2 plot 1) and 27 (BB2 Plot 2) to create a combined chronology of 57 trees (114 cores). Due to core defects (missed pith, compression wood, etc.) some samples could not be used for analysis. From this pool, a subset of 74 individual cores were successfully cross-dated and used for analysis.

44 In addition to the tree cores extracted from each tree, we collected spatial location of each tree using a Trimble Nomad Global Positioning System (GPS) unit. At each point we recorded the DBH of the tree and the density of scale insects using a categorical estimation from 0-5 with 5 having a high density of insects. GPS points were differentially collected and were then uploaded to ArcMap (ESRI 2011) with the accompanying attribute information using Pathfinder Office software (Trimble Navigation Limited) resulting in less than 3 meters of error.

4.3.4 Sample preparation and measurements

All cores were brought back to the laboratory, dried, mounted, and sanded using pro- gressively finer sand paper, up to 600-grit. Each core was visually cross-dated within each tree first using the list method (Yamaguchi 1991). Cores were then measured to 0.001mm precision using a Velmex measuring system (Velmex, East Bloomfield, NY, USA) connected to a Nikon stereo zoom microscope and using J2X software (Voorhees 2000). To ensure ac- curacy in the cross dating method, we statistically verified the dating of the cores using the program COFECHA where ring-widths are used to statistically verify the accurate dating of each core with Spearman’s correlations (with a critical level of 0.3281)(Holmes 1983). The cores were then standardized to remove biological-growth trends associated with tree growth using the package dplR (Bunn 2008) in the statistical program R Version 2.14 (R Development Core Team, 2004). In climatic reconstructions using dendrochronology, it is conventional to remove the effect of tree age, in order to leave a residual ”index” that reflects short-term climatic variation in tree growth (Fritts 2001). In this investigation, however, scale insect density, tree size, and light environment and location are of ecological interest as factors that may influence tree growth. The related patterns of tree growth are ecological processes of interest; therefore the detrending procedures typically used for climatic reconstructions may not be appropriate for this analysis. Therefore for this study we chose a standardization procedure based on the assumption that constant growth is expressed by a constant basal area increment (BAI) distributed over a growing surface (Biondi and Qeadan 2008). BAI removes any age-related growth trend resulting from adding the same volume of wood on an increasing cylinder, while maintaining suppression and release events, due to forest disturbances (Phipps 2005). BAI standardization was performed using the BAI.out function in R package dplR. BAI standardization was performed using the BAI.out function in R package dplR. Sev- eral other functions in the dplR package were useful for data analysis after standardization. These functions included seg.plot (visualize overlap of the series that had been imported), and chron (chronology building), chron.plot (plotting master chronology). For each individual tree in the chronology the mean detrended annual ring width score was calculated. These measurements were then used to uncover any relationships between between tree growth, scale insect density, tree location (edge or forest interior), and size (DBH). Locally weighted regression using the locally weighted scatterplot smoothing func- tion (loess.smooth) in R were performed on variables of interest. Mean detrended ring width scores for each tree were also imported into ArcMap (ESRI 2011) and used to geovisualize tree growth in multivariate space.

45 4.3.5 Pith flecks Speer (2010) describes the ocassional presence of bubbly textured wood called ’pith flecks’ created while aphids suck tree sap and damage the cambium. Structures similar to the photographic representation of pith flecks by Speer (2010) were surveyed in the tree core samples of this study. When pith flecks were observed the growth ring they were present in within tree core sample was recorded. This information was imported to ArcMap to assist in visualizing any patterns to pith fleck abundance in addition to producing plots of individual tree growth over time with presence of pith flecks. In addition to geovisualization, point process patterns for the presence of pith flecks were analyzed. Many forest based questions can be solved through using point processes or marked point processes. The points are considered tree locations and the marks are tree characteristics such as tree diameter or the degree of damage by environmental factors (Law et al. 2009). These point process methods are helpful tools for exploratory data analysis within forestry and are valuable for describing the variability of forest stands and for investigating and quantifying ecological relationships (Stoyan and Penttinen 2000). Data sets in the point process context take the form [xi, yi: mi], given the locations xi =(xi, yi) and the marks mi of all relevant trees in a window W of observation (Diggle 1983). Correlations of the marks in a stationary marked point process can be described by marked correlation functions (Stoyan and Penttinen 2000). The ’spatstat’ package in the statistical program R was used to created point pattern process objects for the two forest areas. Information on tree location was extracted from the GPS coordinates as x and y location and scale fleck presence was used as the marked variable. To further investigate the second order process characteristics the means of the intensity (lambda) were expressed by the Ripley’s K function (Ripley 1976) using the spatstat function ’Kest’ (Baddeley and Turner 2005, Spatstat version 1.25-5). The K function is a test of the hypothesis of complete spatial randomness (CSR)(Illian et al. 2008) and in this research was used to describe the degree of clustering of the presence of pith flecks within sample trees in the forest plot. Resulting observed K values that are larger than the expected K values, indicate the data are clustered rather than being randomly distributed. When the observed value is smaller than the expected K value, the distribution is characteristically more dispersed than a random distribution. A confidence envelope for CSR was created by performing simulations or random permutations in spatstat. Observed K value found to be larger than the high value for the confidence envelope value, indicate that spatial clustering is statistically significant. K values smaller than the lower values of the confidence envelope have statistically significant spatial dispersion or repulsion (Getis 1984).

4.4 Results

Distinguishable ring boundaries were observed after the various stages of sanding the tree core samples. Angiosperms such as Quercus laurina are classified as ring porous with a row of vessels at the beginning of each ring shown in 4.4 where older rings are visible to the left. The differentiation of vessel size at the boundary of each ring (smaller vessels as the tree is slowing growth) assisted in delineating ring boundaries. High elevation cloud forests in Veracruz experience strong seasonality in rainfall and have a distinct period with

46 Figure 4.4: Tree ring boundaries formed in a ring porous oak species, Quercus laurina. Older rings are to the left

substantially reduced precipitation. Climate data from a weather station 10 km from the study area verifies this strong seasonality. A dry season of at least 2 months with less than 50 mm of rain can generate the reduced diameter growth or cambial dormancy to produce annual rings in many species (Worbes, 1999). Reduced rainfall during the months of November-February supports the likelihood of annual ring expression according to this criteria. At site PP, 19 trees (31 cores) were statistically crossdated in COFECHA and were retained in the chronology with an inter series correlation of 0.425, a mean sensitivity of 0.264, and covered the time period 1950-2008( 4.1). Within forest area BB, 28 trees (45 cores) were retained with an inter series correlation of 0.441, a mean sensitivity of 0.293 and covered the time period 1957-2008 4.1. Samples collected at site PP had considerably good sample overlap. Only one tree sample had a period of 10 years (1950-1960) where no other tree could assist in crossdating for those years 4.6. Samples collected at site BB had a different distribution of overlapping years 4.5 with many shorter series with good overlap and a group approximately 7 longer series with moderate overlap. The series intercorrelation is a measure of the stand-level signal of the site and is the average correlation of each series with a master chronology derived from all other series (Speer 2010). Series intercorrelation is a measure of chronology reliability with larger values being more optimal. Most chronologies have values between 0.550 and 0.750, where the

47 Table 4.1: Chronology statistics for all species in all habitats. Series intercorre- lation is a statistic that demonstrates the correlation of each core to the master chronology and is a measure of stand-level signal. Mean sensitivity measures the year-to-year variability and is a measure of sensitivity. The chronology length reports the length of the oldest core.

Forest Area Number of Number Series Mean Chronology Trees of series intercorrelation sensitivity length (years) PP 19 31 .425 .264 58 BB 28 45 .441 .293 51

48 Figure 4.5: The time span and series overlap of each tree core sample collected in 2008 from forest area BB in Chiconquiaco, Mexico

49 Figure 4.6: The time span and series overlap of each tree core sample collected in 2008 from forest area PP in Chiconquiaco, Mexico

50 benchmark for reliable chronologies could be considered above 0.4 (Grissino-Mayer 2001). Trees of some species and in some regions will have higher values than others. The series intercorrelation values for the two sites in this study were within the range of reliable values. Mean sensitivity is a measure of the year-to-year variability in the master chronology (Speer 2010). Mean sensitivity is a measure of inter annual variability in the chronology (ranging from 0-1) with higher values being more indicative of the species ability to detect minor environmental changes (Fritts 2001). A mean sensitivity around 0.1 is too complacent and will be difficult to cross date, while a mean sensitivity greater than 0.4 is too sensitive to date (Speer 2010). Mean sensitivities for site BB and PP were within the range for suitable cross dating. Raw ring-width chronologies show a trend of declining radial growth rates over time (site PP 4.7) (Site BB 4.8). Trees growth rates naturally decrease as they age, so this allometric trend of declining radial growth is expected. The removal of these age-related growth trends through detrending, enables the distinction of radial growth trends unrelated to age, suggesting other external forces (such as climate, forest disturbance, or biotic interactions within the forest) are limiting tree growth. These detrended chronologies (Site PP 4.10 Site BB 4.9) illustrate the ring width values without the progressive decline of ring width along a cross-sectional radial cause by the corresponding increase in stem size and tree age over time. Tree diameter and mean annual growth were positively correlated (R2=0.8716)( 4.11) indicating that larger established trees are growing more than smaller trees. Most den- drochronological studies target trees of similar size classes to minimize these influences, however the objectives of part of this study were to understand the development and regen- eration of younger trees in these forest with varying levels of scale insects, thus the sampling protocol included many small trees. Given this strong size based growth correlation with annual growth, trees were divided into size classes small (5.2-15 DBH), medium (15.5-39 DBH) and large (greater than 39 DBH) when performing additional bivariate plots. Maps created from the attribute data collected at each tree location reveal the same pattern of growth and tree size ( 4.12). Site BB had greater numbers of smaller trees and spacing between these trees was less than the spacing between larger more established trees. All three categorical groupings of tree size classes illustrated the same negative relation- ship between tree growth and scale insect density 4.13. Greater densities of scale insects resulted in lower values of mean annual growth and this relationship was significant for pooled data from forest site BB and PP (p < 0.01; ANOVA test 4.2). Trees with lower densities of scale insects grew better than trees with dense colonies of scale insects. The re- lationship between tree growth and scale insect density varied in relation to the tree position within the forest (edge trees vs. forest interior trees) 4.14 however there was no significant difference in growth between trees located in the forest interior vs. trees located on the forest edge (p > 0.05; ANOVA test 4.2). Trees located in the forest interior hosting fewer scale insects grew better than those hosting more scale insects 4.14. Slow growing trees on the forest edge had a range of scale insect densities whereas faster growing trees tended to have fewer insects 4.14. Due to the relationship between tree size and growth, small edge trees were separated out of the edge tree sample to explore the relationship between tree size, forest position and growth. Smaller trees growing on the forest edge with high densities of scale insects exhibited very little growth 4.14.

51 xxxstdPP RWI SampleDepth 10 15 20 25 30 0 5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 1950 1960 1970 1980 1990 2000 2010 Years

Figure 4.7: The standard chronology is shown for the tree cores collected at site PP using raw ring width data. A smoothing spline highlights low-frequency variability and the sample depth (number of tree cores used in the chronology for that time period) is plotted on the right-hand y-axis with a dotted line

Table 4.2: Analysis of variance results of tree growth as a function of scale insect density, tree size and location (forest edge or interior). Data are pooled from both forest areas (BB and PP).

Df SumSq MeanSq Fvalue Pr(>F) SCALE 1 16805150.85 16805150.85 29.62 0.0000 DBH 1 219841183.49 219841183.49 387.53 0.0000 POSITION 1 214251.84 214251.84 0.38 0.5422 Residuals 42 23826026.63 567286.35

52 BB xxxstd RWI SampleDepth 10 20 30 40 2 3 4 5 0 1960 1970 1980 1990 2000 Years

Figure 4.8: The standard chronology is shown for the tree cores collected at site BB using raw ring width data. A smoothing spline highlights low-frequency variability and the sample depth (number of tree cores used in the chronology for that time period) is plotted on the right-hand y-axis with a dotted line

Table 4.3: Analysis of covariance results of tree growth as a function of tree size and scale insect density as the covariate. Data are pooled from both forest areas (BB and PP).

Df SumSq MeanSq Fvalue Pr(>F) SCALE 1 16805150.85 16805150.85 30.77 0.0000 DBH 1 219841183.49 219841183.49 402.50 0.0000 SCALE:DBH 1 1100454.66 1100454.66 2.01 0.1632 Residuals 42 22939823.81 546186.28

53 xxxstdBB BAI RWI SampleDepth 10 20 30 40 1000 2000 3000 4000 5000 6000 0 0 1960 1970 1980 1990 2000 Years

Figure 4.9: Detrended chronology for the tree cores collected at site BB using basal area index data. A smoothing spline highlights low-frequency variability and the sample depth (number of tree cores used in the chronology for that time period) is plotted on the right-hand y-axis with a dotted line

Pith flecks 4.15 were observed in 28 of the sample trees. Site PP had 10 trees with typically 1-2 annual growth increments with visable flects. Two trees at site PP had 3 and 7 annual growth increments with Pith flecking. Site BB had 18 sample trees with scale flecks found within 1-2 annual growth increments. Approximately 80% of all the trees with scale flecks had noticeable flecks on both sample cores taken at each tree, suggesting that when scale flecks are present they are consistently found at various directions around the circumference of the tree. When annual growth in individual trees was plotted in association with years of pith flecking there was no noticeable pattern to years with greater or lesser growth increment 4.16. Locations trees with pith flecks appear to be clustered in location 4.17. Statistical tests to investigate clustering resulted in K values that were larger than expected K values. Observed K values were found to be larger than the high value for the confidence envelope value, indicate that spatial clustering is statistically significant and does not fit the model of complete spatial randomness (CSR). 4.18

54 xxxstdPP BAI RWI SampleDepth 10 15 20 25 30 1000 2000 3000 4000 5000 6000 0 5 1950 1960 1970 1980 1990 2000 2010 Years

Figure 4.10: Detrended chronology for the tree cores collected at site PP using basal area index data. A smoothing spline highlights low-frequency variability and the sample depth (number of tree cores used in the chronology for that time period) is plotted on the right-hand y-axis with a dotted line

4.5 Discussion

Most investigation on scale insects have focused on morphology, pest status in agricul- ture situations, or role in trophic interactions with ants. Few studies have examined the effects that scale insects have on the growth of trees within forest communities. The often sometimes cryptic appearance or bark burrowing habit of scale insects makes the ecological effects on trees not readily apparent. Herbivory by scale insects in conjuction with habitat variables in this study were shown to affect the forest growth dynamics in Chiconquiaco, Mexico. In this study we find support that the interaction is influenced by the costs of chronic herbivory (oak tree growth declines with increasing scale insect density supporting(H1). We found little that this association is influenced by the habitat selection decisions of ovipositing female scale insects where oak tree growth and scale insect density are positively correlated due to the positive effects

55 Figure 4.11: Mean basal area growth plotted as a function of tree diameter for all sample trees in Chiconquiaco, Mexico

of high-light environments of both scale insects oviposition density and tree growth (H2). Much of this finding may be influenced by the fact that within our sampling scheme, edge trees were usually smaller in size and the detriment of scale insects may override the benefit of growing in a high light environment. The finding that greater densities of scale insects resulted in lower values of mean annual growth even in larger established trees was somewhat surprising. One would expect that larger trees may be able to compensate for the removal of some photosynthetic material, however it appears that even in larger trees with high densities of scale insects, the optimal growth potential of these trees is compromised. The rejection of (H2) (negative response to herbivory is offset by increased sunlight) may be due to microclimatic conditions outside the range of what is typical for a species of oak tree that is suited for growth in a montane cloud forest (ie high light, less humid microclimate may actually be more stress inducing for these oak trees). In addition populations, communities or ecosystems with an evolutionary history of environmental stability may be most affected by perturbations (Clark 1991) Slow growing trees on the forest edge did however have a range of scale insect densities

56 Mean Growth and Tree Size

Site BB Site PP

Mean growth DBH 0 - 8 136 - 384 9 - 26 385 - 802 27 - 49 803 - 1371 50 - 62 1372 - 2100 2101 - 4841 63 - 108 4842 - 8612

Figure 4.12: Tree locations at both study sites (BB, PP) mapped with increasing symbol size as a function of increasing tree diameter and with color categories indicating annual mean growth

57 Figure 4.13: Locally weighted regression of tree growth as a function of scale insect density and tree size classes (small diameter (5.2-15 cm), medium diameter (15.5-39 cm), and large diameter (> 39 cm)) using the locally weighted scatterplot smoothing function. Data were combined from forest site BB and PP.

58 Figure 4.14: Locally weighted regression of tree growth as a function of scale insect density and location (forest edge, forest interior) and location and size (small trees on the forest edge) using the locally weighted scatterplot smoothing function. Data were combined from forest site BB and PP.

59 Figure 4.15: Small areas of wood resembling tree pith (’pith flecks’) are caused by damage to tree cambium by the feeding of scale insect, Stigmacoccus garmilleri. These pith flecks are visible and highlighted here within prepared tree cores from Quercus laurina in Chiconquaico, Mexico.

60 Figure 4.16: Individual years pith flecks were observed (Forest area BB, tree 22) plotted upon yearly growth increment values for the lifespan of the tree.

61 Scale Density and Presence of Pith Flecks

Site BB Site PP

Scale density Pith flecks 0 1 1 2 2 3 3 4 5 4 6 - 7 5

Figure 4.17: Map illustrating the relationship between tree core sample location, scale insect density, and numerical occurence of pith flecks within core samples for each tree. Scale insect density is illustrated with graduated colors while pith flecks are represented by graduated symbols.

62 Figure 4.18: Degree of clustering of trees with observed pith flecks plotted by Ripley’s K function tested against a confidence envelope for complete spatial ran- domness (CSR) created by performing simulations or random permutations.

63 and some trees in the forest interior maintained lower values of mean growth than similarly sized forest edge species, indicating that there may actually be some benefit to tree growth being located on the edge. The forest structure in site PP actually allows more incident sunlight to reach the forest interior trees due to the large spacing between individual trees with little undergrowth due to grazing, which may have masked any strong interior - edge growth relationships. The finding that smaller trees growing on the forest edge with high densities of scale insects exhibit very little growth is concerning for the future of these forest remnants. Reforestation procedures in Mexico often concentrate on planting seedlings at the edge of forest patches, and in the case of the montane oak forest of Mexico this success of this procedure will be complicated by the fact that the small edge trees are quickly colonized by scale insects from the dense distribution of these insects on forest edge trees. This is the first study that has quantified and mapped the evidence of pith flecks. The finding that pith flecks were found in both tree core samples at individual trees should support further studies using tree core sampling to get an accurate account of the occurence of pith flecks without having to prepare cross sections of the trees. Further investigation in the pattern to years in which the flecks are formed would be valuable. Analysis of spatial point patterns is a subject of current research in statistics (Illian et al. 2008). The type of ecological insight that can be gained from point process theory are best seen in the context of specific ecological issues (Law et al. 2009).The finding of a statistically significant clustered spatial distribution of pith flecks by point process analysis leads insight into the dispersal capabilities of these scale insects. Although the males are winged and allow for movement or gene transfer, female insects are likely bound to the nearby environment in which they can crawl to during the short time period between mating and ovipositing in the oak trees. Information shed on the dispersal and spread of these insects could be applied to adapt a better reforestation strategy such as the inclusion of a zone of no planting where seedlings could be placed outside the dispersal limitation of the ovipositing female insects.

64 CHAPTER 5

A SOCIOECOLOGICAL CRITIQUE OF BEEKEEPING AS MECHANISM TO OFFSET DEFORESTATION: THE IMPORTANCE OF CONTEXT FOR HONEY PRODUCTION IN RURAL VERACRUZ, MEXICO

5.1 Abstract

Much of the tropical and subtropical landscape today is a matrix forest fragments sur- rounded by grazing pasture or planted crops. Honey bee forage (flower nectar or plant exudates) for the production of honey can be found throughout the matrix, although in some areas, production of honey is greatest where bees can access the remaining forest frag- ments and produce ’forest honey’. Thus, beekeeping has been promoted as a livelihood that can promote the production of non-timber forest products (NTFPs) and lessen dependence upon livestock grazing, a primary contributor to landscape fragmentation. Beekeeping can benefit the local community by generating income through the sale of honey and in some areas through fees charged to providing agricultural pollination services. The reality of beekeeping today however introduces a nuanced set of socioecological issues that can com- plicate the chances for its long-term implementation. Vegetation change associated with transitions in land use, the social and economic networks involved in beekeeping, bee dis- eases and pests, and ’Africanization’ of European honey bees necessitate consideration when implementing a beekeeping project to ensure its success. To illustrate the arguments in our critique, we describe the potential for beekeeping in the fragmented montane oak forests of Mexico. We examine the socioeconomic and ecological context of deforestation and the promotion of beekeeping for the township of Chiconquiaco in Veracruz State. We discuss the potential for these fragmented forested areas to support apiculture to lessen dependence upon the agricultural land uses that are driving forest fragmentation. Land use change in the area causes changes in bee forage that may be correlated with the microclimatic shifts accompanying fragmentation. Our critical examination of apicultural development and our case study in Veracruz serve as a template for a more adaptive framework for implementing forest beekeeping projects.

65 5.2 Introduction

Humid tropical forest habitat is declining globally from climate change (Loarie et al. 2009) and by human land use change (Hansen et al. 2008). After trees are removed for timber, land is commonly converted to other land-uses such as cattle pasture or agricultural fields. Although tropical forest conservation is a top priority for human and environmental health, deforestation persists, mainly because of food and economic needs. Individuals as well as communities may be reluctant to give up economic activities for the sake of ecological integrity, unless alternative economic activities are presented. In response, a variety of mechanisms have been put into place to slow the incentives to clear forest. Forest certification programs such as the Forest Stewardship Council provide internationally recognized standard-setting accreditation services to companies, organiza- tions, and communities interested in responsible forestry (Eden 2009, Klooster 2010). Inter- national climate change mitigation initiatives such as Reduced Emissions from Deforestation and Forest Degradation (REDD) may promote the retention of forest cover and clean de- velopment mechanisms (Sandbrook et al. 2010). Community forestry can also be a viable approach to forest conservation and community development. With de-centralization and greater local participation, communities have more control of resources and benefits from those resources (Charnley and Poe 2007, Kainer et al 2009). In addition, community- based ecotourism may provide economic incentives to resist or slow deforestation (Kiss 2004). The promotion of non-timber forest products can also add value to forests and make them more robust to conversion (Garcia-Fernandez et al. 2008). Beekeeping and the pro- duction of ’forest honey’ is often promoted as a non-timber forest product and one method to slow the incursion of deforestation, and to promote livelihoods that lessen incentives to clear forest. In Mayan communities of Mexico, a long history of stingless beekeeping (meliponiculture) supports cultural preservation and forest conservation (Chemas and Rico Gray 1991, Portar Bolland 2003). Rural beekeeping projects are also described from Brazil (Brown 2001) and Africa (Illgner et al. 1988, Ingram and Njikeu 2001, Sande et al. 2009). From the view of the general public, beekeeping can be perceived as a form of economic development that can be deployed wherever there is vegetation. Although it is indeed a very common cultural practice, its promotion and success belies more complexity. Like community forestry, carbon offset mechanisms, and ecotourism, there is an on-the-ground complexity that is often overlooked and merits scrutiny. In this paper, we put forward that the deforestation-beekeeping linkage needs to be more critically examined in order to improve on its effectiveness. We propose several arguments as critique of beekeeping as static enterprise that can be delivered as relief aid. Although this type of critique is longstanding in the development literature (Ingram and Njikeu 2011), it has not been as thoroughly examined for apiculture. It is a truism to state that the transfer of knowledge and skills must accompany an apicultural project for it to be sustainable. What is less considered is the importance of the context. Without consideration of the underlying heterogeneity in the social and ecological systems that underlie beekeeping, its promotion is not likely to persist as a livelihood. Bees and their beekeepers are heavily dependent upon climate and the availability of floral resources. Deforestation involves changes in pollen and nectar resources that may impact beekeeping in positive, negative, and sometimes contradictory ways. Different types of bees can also be used to produce

66 honey. The type of honey bee managed by the beekeeper has important implications for the quantity of honey that can be produced, the way honey is collected, and the amount of human labor required in tending colonies. In addition, the variety and type of floral sources available to beekeepers will influence the supporting market potential for honey. Forest honey varieties can be prized in one culture yet disdained in another. Implementation of beekeeping requires a social context that will promote the training and skills to see it succeed. A wide range of people are involved in beekeeping, not just the beekeepers. From the craftsmen who make and sell beekeeping tools, to landowners and farmers who may benefit from proving access to floral resources, a network of interactions arise not only in an ecological sense but also within a social context. These socioecological networks that form around beekeeping enterprises can be expected to vary in time and space. The first half of this paper elaborates on these issues to illustrate more of the complexity inherent in beekeeping. Although division is artificial, we first discuss the ecological facets of initiating a beekeeping project, followed by a presentation of the more social facets of relevance. The second half of the paper provides an example of how this more critical per- spective plays out for a rural location in Veracruz, Mexico. Deforestation and fragmentation of humid tropical montane oak forests in Veracruz have greatly reduced forest cover, and mechanisms are needed to lessen incentives to clear forests and convert land to agriculture and grazing. Through our Veracruz case study we show how a one-size fits all approach to beekeeping would, in our case study, ignore many of the contingencies and opportunities that would be central to its success. Our critique is not meant to lessen the promotion of beekeeping as a productive conservation strategy. On the contrary, our aim is to increase awareness of more of the nuances of beekeeping, and to promote a more robust framework for planning a beekeeping project. Our goal is to make beekeeping more sustainable, from the perspective of offsetting deforestation, but also in terms of the residents who invest their time and labor. In sum, we are outlining an adaptive approach to apiculture, one that can lead to more investment in the long-term resiliency of community-driven responses to deforestation.

5.3 Ecological Knowledge

Studies on gaining income while maintaining forest cover (productive forest conservation (Hall 1997)) are often anchored within the realm of environmental politics. Apiculture as a productive conservation strategy may consequently forego or postpone explicit charac- terization of the ecological and biological context (Brown 2001). However, comprehensive biological knowledge to sustain colonies of honey bees may not be by itself sufficient for a successfully implemented and sustained forest conservation project. An even broader ecological view of apiculture is required to comprehend its sociopolitical entrenchment. If apiculture is to be a mechanism to promote productive conservation, then it is likely to be undertaken in landscapes where the conservation of other biota is prioritized. Thus an overarching ecological context for any beekeeping initiative is the status of local ecological dynamics related to land use change, landscape configuration and its influence on native pollinators and other keystone, umbrella, or ecosystem engineering species (Wilcox 1984,

67 Jones et al. 1994, Paine 1995).

5.3.1 Land-use change and landscape configuration

Land use change is a major driver of native pollinator declines at global scales (Holzschuh et al. 2011, Garibaldi et al. 2011). Beekeeping initiatives will likely take place in modified landscapes that influence native pollinator abundance and distribution. Where ecosystems are fragmented, the richness and abundance of native pollinators can decline (Aizen and Feinsinger 2003). Flower-insect interactions are negatively affected by forest isolation (Kre- men et al. 2004, Brosi 2009). However, the magnitude of these changes on native pollinator interactions will vary based upon on the type and intensity of land use change. Shortages of flower resources in landscapes of lower complexity, may affect the reproductive success of native pollinator colonies negatively (Persson and Smith 2011). The most significant negative affect to pollinators can be expected in areas where little native habitat remains (Winfree et al. 2009). Fragmentation may not always have negative effects on native bee communities in terms of density and diversity (Becker et al. 1991, Cane 2001). Moderate levels of fragmentation may promote coexistence of pollinators (Winfree et al 2009). The edges of forest patches in fragmented areas allow bees to exploit the greater temporal and spatial diversity of floral resources both inside and outside the forest (Chacoff and Aizen 2005). It has also been shown that remnant forest areas which are homes to a variety of pollinators, can increase crop yields in neighboring agricultural areas (Chacoff and Aizen 2006). Beekeeping is complexly intertwined with vegetation dynamics when one considers the role of honey bees as pollinators. In a beekeeping project, questions may arise as to how domesticated honey bees could influence the success of one plant taxon over another. Do- mesticated honey bees require large amounts of floral resources and collect these resources as efficiently as possible (Seeley 1985). Thus, they tend to visit flowers which have rich floral rewards. Honey bees typically visit flowering species in larger resource patches while ignoring the least abundant flowers (Butz Huryn 1995). These landscape and ecosystem-level relationships have relevance to the design of a beekeeping project. Whereas edge and fragmentation are important from the view of native pollinators, it is also important for domestic honey bees. In a conservation area in Kenya, Sande et al. (2009) studied honey yield from managed honeybee hives placed at varying distances from the forest reserve. Honey yield almost doubled in hives placed less than one kilometer from the forest compared to those placed more than three kilometers from the forest. Such remnant forest areas can provide target habitats for maintaining colonies of honeybees and for subsequent economic maintenance of rural communities that live amongst these areas. The strength of this edge effect will likely vary in response to compositional differences of the surrounding landscape vegetation. Placement of hives of domesticated honeybees might be designed in a way to offer varying levels of integration of native and domestic honey bees. This may circumvent some of the polarizing issues regarding the potential negative interactions between native pollinators and introduced honey bees.

68 5.3.2 Non-native pollinators

Perhaps one of the greatest challenges of introducing a beekeeping project centers on the perception of their degree of nativeness. Even though feral or managed colonies may have long been present at a given location, domesticated honey bees often retain the status as non-native and introduced. For that reason, a beekeeping project should consider the local debates and opinions on the competitive outcomes of native and non-native domesticated pollinators. The certainty of any particular outcome from the introduction of domestic bees is at times ambiguous; however any negative impacts of introduced bees need to be carefully assessed before further introductions are carried out. A global-scale review study finds that honeybee invasions seem to have had little if any effect on biodiversity of native pollinators (Moritz et al. 2005). Other studies suggest that honey bees compete with native pollinators for floral resources (Roubik 1980, Schaffer et al. 1983, Steffan-Dewenter and Tscharntke 2000, Thompson 2006). In areas of concentrated honey bee foraging, bumble bee workers had reduced worker size, perhaps due to compe- tition (Goulson and Sparrow 2009). In addition to these effects it has been shown that pests and pathogens found in European honeybees are transmitted to native bees (Kojima et al. 2011). Beekeepers monitor and treat for these maladies; however, the effects on the population dynamics of untreated wild bee populations are unknown. Further impacts of non-native bees include decreased pollen transfer per visit because of the loose morphological correspondence between flowers and non-native bees as well as changes in pollination efficiency caused by non-native pollinator behavioral traits (such as stealing nectar or collecting previously-deposited pollen from stigmas)(Dohzono and Yokoyama 2010). In addition, changes in climate and disturbance regime, either histor- ical or in the near future, may alter these interactions (Burkle and Alarcon 2011). Climate change and non-native species can lead to the creation of novel communities; however, climate change and alien species in combination can result in significant threats to more specialist interactions involving native species (Schweiger et al. 2010). Exotic bees also often exhibit marked preferences for visiting flowers of exotic plants and are often primary pollinators for a number of exotic pest plants (Goulson 2003). Introduced bee species were shown to be superior pollinators of exotic plants in comparison to native bees (Gross and Mackay 1998, Barthell et al. 2001, Goulson 2003, Liu and Pemberton 2009). Invasive bee pollinators can also disrupt mutualistic plant-animal interactions (Barthell et al. 2001), and thus affect native plant seed production (Gross and Mackay 1998). In montane tropical rainforests of Australia the location of colonies of managed honeybees is a controversial subject after finding that native bees were disturbed from foraging at flowers and seed set was significantly lower in a native plant species with honey bee pollinators (Gross and Mackay 1998). In Florida, domestic honey bees depend upon the invasive shrub Brazilian pepper for forage during times of the year when other plant sources are unavailable. Despite the potential harm exotic species can cause to biodiversity, some dispute the current global campaigns against exotic species since they are often not context dependent and are solely based on biological arguments (Prevot-Julliard et al. 2011) In regards to decision making for introducing beekeeping to natural areas we plea in favor of context- based discourses and decisions, based on both ecological knowledge, dialogue and local conservation objectives.

69 5.3.3 Selecting a bee species The type of bee is a factor to consider when initiating beekeeping as a form of produc- tive conservation. An alternative way to increase the diversity of pollinators is to manage different types of bees. Some native bees can be domesticated with varying levels of success, as with stingless beekeeping (Villanueva-Gutirrrez et al. 2005) and bumble bees (Roseler 1985). Beekeeping using stingless bees of the family Meliponidae is known as meliponicul- ture. Stingless beekeeping is practiced in tropical regions where stingless bees are native. Meliponiculture in Mexico is considered an ancient tradition that is in decline and near disappearing as a result from changes in cultural, economic and ecological factors (Quezada- Euan et al. 2001). Stingless beekeeping reached an impressive level in the Yucatan penin- sula, where Mayan people developed this form of beekeeping to a similar scale as that of honeybees in Medieval Europe (Labougle and Zozaya 1986). The highlands of Puebla and Veracruz State, Mexico, meliponiculture reached more modest development where ethnic groups such as the Nahua, Totonaca, and Chontal kept stingless bees (Quezada-Euan et al. 2001). Typically stingless bees produce smaller quantities of honey than honey bees (Apis spp.) and the honey is considered more medicinal in its purpose (Villanueva-Gutirrrez et al. 2005). The European honey bee, Apis mellifera, and several of its popular subspecies (Ital- ian honeybees (Apis mellifera linguistica), Carnolian honeybees (Apis mellifera carnica), Russian honeybees (variant of Apis mellifera carnica)) are commonly bred and managed throughout the Americas. African honeybees (Apis mellifera scutellata), were introduced into Brasil in 1956 for experiments with honey production and they have since escaped and prospered throughout South America, Central America, Mexico and are now found in several states in the United States (Pinto et al. 2005). European honey bees now kept in Latin America are mostly ’Africanized’ and these hybrid bees are known to be very de- fensive of their hives, which requires beekeepers managing them to take extra precautions such as using heavy protective equipment and placing colonies away from dwellings or do- mestic animals (Guzman-Novoa and Page 1994). These management considerations should guide decision making on selecting a bee species. Often the increased honey production of Apis mellifera in comparison to species of Meliponidae will dominate this decision toward using honey bees; however the benefits of reviving an ancient tradition and production of medicinal honey with minimal concern for safety in the rural community should not be overshadowed by production limitation.

5.3.4 Access to floral resources Access to floral resources is critical for the success of a beekeeper. Floral resources can vary greatly over small distances and over the course of a year. Although bees can forage up to distances of 9 km from the hive (Beekman and Ratnieks 2001) they can maximize their efforts by foraging closer and are thus very sensitive to localized heterogeneity in floral resources. Even within a single land use, the idiosyncratic occurrence of groups of flowering trees or shrubs can provide a food source. However, the presence of flowers alone does not imply an adequate source of forage for a beekeeping initiative. Domesticated honey bees select forage based on nectar concentration and sugar composition as well as abundance. In addition flower pollen is required as a source of protein. Not all pollens are equal in

70 terms of their amino acid composition, and bees will require resources that correspond to their metabolic needs. Determining whether a habitat is suitable for bees is a trial and error process. Understanding of this suitability cannot be immediately gauged unless local apicultural knowledge is available. A variety of habitats may promote beekeeping, even locations that have a significant human impress. Pristine habitat may actually have reduced forage when compared to areas where some human activity has introduced a greater and more variable cover of ruderals and cultivars and perhaps exotic invasive plants (Levy 2011). Thus it may not be necessary to place bees deep within conserved areas. Even suitable habitat types for maintaining honey bees will experience quality differences driven by temporal climatic changes like the Southern Oscillation or the North Atlantic Oscillation especially in regards to honey production (Maxwell and Knapp 2011).

5.3.5 Mobility of Beekeeping By relocating hives throughout the course of a year (migratory beekeeping) beekeepers can provide a wider selection of forage. When bees cannot be moved to take advantage of changing locations of floral sources, dependence upon the local setting may be acute. If no forage is available locally or at a distance, bees have to be supplemented with a carbohydrate source, typically sugar water or corn syrup, and also be fed bulk pollen substitute to assist in brood rearing. This is an added expense for beekeepers, and may be prohibitively so. Moreover, beekeepers are often pressured to remove all honey stores within the hive in order to make enough profit through the sale of honey. Robbing bees of their honey or pollen only increases the colony’s dependence upon floral resources as well as the practices of the beekeeper to safeguard against starvation.

5.4 Socioecological context

5.4.1 Social networks Awareness of social, cultural, and economic context will aid in the development and implementation of a rural beekeeping program. Several established organizations help dis- tribute information to those interested developing community beekeeping programs (e.g. Bees for Development and Heifer International). A large promotional literature supports beekeeping as an ideal small farm development strategy, often in the forefront touting that beekeeping requires relatively few labor and capital inputs (Lalika and Machangu 2008), and does not interfere with already existing farm practices (Brown 2001). However, livelihoods, regulatory framework, business environment and forest management are intricately linked in the chain from beekeeper to consumer. This makes it difficult for beekeeping to be a direct route out of poverty (Ingram and Njikeu 2011). A common trend in written reports regard- ing linking beekeeping to rural livelihoods is the initial discussion of a beekeeping ’utopia’. Only later in the study, when conclusions and future project considerations are drawn up is there sufficient attention to the relevance of social factors. Much of the final success of rural beekeeping will lie in the awareness of the socioeconomic network, and context of this network prior to project implementation. This network may be small but it needs to be in

71 place to ensure success. Providing beekeepers with equipment and appropriate techniques will not automatically make beekeeping successful and move people out of poverty.

5.4.2 Addressing economic opportunity Beekeeping activities should mitigate the impact of people’s poverty on the natural environment. They should be promoted along with the improvement of other livelihood sources. If beekeeping is to be used as one of the tools for poverty alleviation, it is important to ensure that the most vulnerable have access to the market. One of the new forms of intervention arising from the failures of modernization is the relocation of control to local stakeholders. Control can be vested in the host community and thus draw from local skills and resources to improve living conditions (Illgner et al. 1999). Beekeeping is an activity around which such community cooperatives can form. Beekeeping lends itself to cooperative action in collecting honey and purchasing or manufacturing equipment and offers great opportunities for both co-operation and self reliance, as it is normally more successful when practiced collectively (Brown 2001). Moreover, to generate sufficient income to make a difference, beekeeping has to persist on a sufficiently large scale (Bees for Development 2011). A larger number of people are required even beyond those who directly maintain hives. Upon initiation of the beekeeping project willing participants should be organized and empowered to gain access to this knowledge and resources. An example of these considerations includes but is not limited to:

1. Preventing absentee landowners from taking a quick profit from forest felling

2. Showing participants how to make inexpensive hives from local materials

3. Creating additional employment opportunities to craftsmen who can manufacture gloves, veils and other beekeeping equipment

4. Organization of training sessions on the basics of apiculture from nearby apiculturalists

5. Organization of more advanced training sessions on pests and pathogens from nearby academic institutions

6. Development of a community based honey extraction, bottling and storage facility

7. Creation of wax based processing and product creation (candles, cosmetics)

8. Exploring additional economic gains that can be related to pollination services pro- vided to nearby agricultural land owners

9. Forming a marketing team from within the community of beekeepers

One major consideration is that too much economic support of a beekeeping develop- ment project at the start can reduce the long term potential for sustaining the project. Once project grants end, it will be extremely difficult for local cooperatives to continue providing the essential services of teaching farmers how to keep bees and processing, pack- aging and marketing the honey farmers produce (Brown 2001). By neglecting relatively

72 simple economic and ecological considerations in the implementation of projects, produc- tive conservation projects may unwittingly cripple the efforts of small farmer organizations to establish sustainable livelihood options in this tropical rain forest region. Indigenous knowledge may enable communities to carry out beekeeping activities at minimal cost, since it often requires less financial investment and capital especially for the construction of equipment. Nevertheless, in terms of production, indigenous knowledge has a positive impact on the quantity and quality of bee products that can be produced. Before initiation of any beekeeping project a general survey of skill availability should be under- taken in conjunction with a needs assessment to determine if the skills are currently available or whether there is local knowledge from other beekeepers or a nearby training program in place. Local beekeepers will have a depth of valuable and applied practical knowledge that if available needs to be incorporated into the development of the new beekeeping program. Foreign technical expertise is often theoretical in nature and is more difficult to integrate into practical work. Using this foreign theoretical knowledge often leads to the use of ex- pensive or inappropriate solutions in the face of abandonment of well tried and tested ideas under the assumptions that new ideas are better ideas (Bradbear 2009).

5.4.3 Gender roles - women participants Beekeeping can also lead to the empowerment of women within a community. In the Kaempa district of Zambia honey hunting is a traditionally a male dominated activity, but in recent years female heads of house hold have begun to keep bees (Illner et al. 1998). Bee- keeping can therefore encourage gender empowerment, allowing women to gain an improved status and community role through their capacity to generate an income. Beekeeping ini- tiatives in Calakmul, Mexico found women to be extremely capable of managing bees in addition to making the most of the products of the hive (honey, wax, and propolis). Training in Calakmul is given to women in the use of phone and fax, making business contacts, and sales so that the women themselves can undertake more roles in the system (Aguiree and Pasteur 1998). These enhanced roles can provide a new endeavor that is uniquely run and organized by women groups within communities and help ensure longer term sustainabil- ity of beekeeping activities. This structure would involve training of women in beekeeping management, and processing and marketing of bee products in addition to the encourage- ment of equal participation of women at all levels of beekeeping development interventions (Ahmad et al. 2002; Bradbear et al. 2002; Adgaba et al. 2008; Horn 2011).

5.4.4 Negotiating access to forage Access to forage and ownership issues in many rural communities can be complicated by the fact that many of the landowners live in more affluent cities and those present on the land live in modest establishments while tending livestock or cropland. Prior to the program establishment it must be determined how profits from the sale of honey, will be distributed between the beekeepers and the landowners. The determination of individual hive placement by beekeepers can be a contentious issue and pre-planning in this regard may head off any concerns. When competition for sites is high, beekeepers can sometimes sabotage or steal hives of other beekeepers. These possibilities can be lessened through the formation of well functioning cooperative with an open and clearly established economic

73 structure. A well functioning cooperative may serve to minimize any negative internal politics amongst individual beekeepers. When shaping this structure policies should be set in place as to address newcomers who may arrive to the project, particularly if the beekeeping project has been successful.

5.4.5 Pests, pathogens and africanized bees Many pests and pathogens plague bees and beekeeping. The hive environment is a desirable location given that it is maintained at a stable temperature throughout the year and has copious food reserves at times. Both developing and adult bees serve as hosts and vectors for parasitic mites, viruses, bacteria, fungal pathogens in addition to other microbes (Evans and Schwarz 2011). Successful management of colonies today often requires knowledge of much more than just basic honey bee biology. A beekeeper must understanding the population dynamics of the parasitic Varroa mite in addition to understanding key features to the other dominant microbial issues. To complicate matters further, new novel pathogens (Runckel et al. 2011), are currently being discovered, and beekeepers need to have access to current information and resources necessary to control them. Certain races of honeybees are superior to others in regards to pest resistance; although often this benefit may be at a cost to honey production or another desirable colony trait, such as gentleness. Africanized bees are known to be much more pest resistant than European races of honey bees (Moretto and Mello 2000). Colony Collapse Disorder (Johnson 2010) has been confined to largely North American and European settings where an Africanization of European honey bee has not occurred. This pattern may support the idea that Africanized honey bees are more resistant to pests and pathogens. Beekeepers in Latin America can benefit from the higher pest resistant traits of Africanized bees; however a more detailed cost-benefit analysis needs to be completed to understand the how other traits of Africanized bees (honey production, lost bees to swarming, and defensiveness) factor in.

5.4.6 Challenges to marketing honey In many farm projects with rural communities the transportation and storage difficulties, can work against the attempts of small-scale farmers to compete as suppliers of food to external markets. However honey and beeswax are a high-value commodity that are easy to transport and store and can provide a valuable source of income. Honey does not spoil if properly harvested and processed and it already has well-established local, regional, and international markets. Wax is an important by product of beekeeping that can be melted down and stored for many years. Honey however if not marketed or consumed, may eventually start to ferment. The quality of extracted honey, water content and temperature at which it is stored will determine the time frame at which this will occur. It is seldom worthwhile to tap into global markets if an acceptable local market is avail- able (Bees for Development 2011). The commodity prices for honey are not usually high enough to be more profitable than local or regional markets, especially once any invest- ment costs have been subtracted. The costs of complying with foreign regulations can be significant and selling honey to foreign markets may only be cost effective if specialist Fair Trade or Certified Organic marketing schemes were accessed, however there is a rich liter- ature regarding the establishment of fair trade mechanisms and the problems that arise as

74 a result. As shown with the example of coffee, fair trade can be an exclusionary practice, leading to favored status of some groups and the exclusion of others (Mutersbaugh 2009). Before entering on this kind of investment it would be first recommended establish a local market or to find a reliable local buyer. It may be more beneficial to investigate expand- ing local markets through acquisition of possible supermarket contract or other high value stores. These may be worth investigating through local business people with connections into these communities. In addition, beekeeping gives local people and the government economic incentive for the retention of natural habitats and forest conservation economic subsidy programs (Lalika and Machangu 2008). Entry into any large scale market will require a means of ensuring high quality honey free form dirt, contamination or adulteration while small quantities of honey will need bulking or consolidation of honey into larger quantities so a reliable supply can be made available throughout the year. A reliable supplier offering high quality honey that they can ensure is pure, unadultered and clean will secure a solid customer base who frequently return to buy more honey. Development partners should fund research to ascertain ways of improving harvesting and marketing of bee products to enhance their quality and quantity and hence improve the livelihoods of people in the beekeeping community. These partners need to consider several important issues while considering the economic and logistical needs of the beekeeping community. The initial cost of the equipment needed to start understanding of how long this will last before replacement or additional equipment is needed, should be predicted. Following this estimate, whether or not these initial costs can be fulfilled without borrowing money needs to be assessed. Borrowing money for beekeeping is rarely economically beneficial and can leave some producers in greater poverty than they started if something were to not work as planned. After local economic assessment understanding the surrounding honey market is fun- damental and vital. The structure of this market and knowledge of buyers and price by weight will be starting points to factor into economic decisions. Understanding the local uses of honey (e.g. used predominantly as a high energy food or valued more for treatment for coughs, skin infections and other common ailments?) needs to be assessed in order to target the type of honey that is produced or advertised. Following this the scale of produc- tion should be evaluated and measured to meet the income aspirations given the market conditions. If the production at this scale outweighs the local demand then perhaps wider markets need to be sought. Lastly, additional income that can be made from the production of secondary hive products such as candles and cosmetics, pollination services to farmers, beekeeping equipment, and selling colonies of bees can be factored in once the base honey production evaluations are made. A complete business plan should be written down and in place and understood by all participants before deciding to go ahead with each of these individual steps.

5.4.7 Non-floral resources Honey is a complex mixture produced by honeybees from the nectar and also exudates from plants and it is consumed as a sweetener as well as for its therapeutic properties (Rodriguez et al. 2012). Monofloral honeys are specific types of honey with a distinctive flavor, color or aroma due to the origin of nectar from one plant species. In general, honeys

75 classified as monofloral are much more appreciated commercially, which are reflected in the national and international markets (Villanueva-Gutierrez et al. 2009). In contrast to multifloral honey where there is no dominance of any pollen type in the sample, monofloral honey is identified by the presence of one plant pollen type that dominates 45% or more of the pollen in a sample of around 300 pollen grains (Louveaux et al. 1978, Dustmann 1993, Molan 1998). In addition to the pollen characteristics, the composition of botanical com- ponents permits the verification of the authenticity of the honey. The nectar of each plant species gives the honey characteristic aroma, color, taste, and physio-chemical properties (Molan 1998) 5.1.

Figure 5.1: Market display of the variety of honey colors and types produced by beekeepers from the state of Veracruz, Mexico

Some common examples of monofloral honey from North America include clover (Tri- folium spp.), orange blossom (Citrus spp.), tupelo (Nyssa ogechee), sage (Salvia spp.), buckwheat (Fagopyrum spp.) and sourwood (Oxydendrum arboreum). In Mexico and Cen- tral America few monofloral species have been described, despite their possible abundance. Recent studies in Mexico have described Eucalyptus (Eucalyptus globules), orange blossom (Citrus spp.), bell flower (Campanula persicifolia), and cactus flower (Opuntia ficus), in addition to several monofloral honeys from the Yucatan peninsula in Mexico (Villanueva- Gutirrez et al. 2009, Rodriguez et al. 2012). Both eucalyptus and orange blossom honey were noted to be good sources of antimicrobial and antioxidant compounds that might main- tain good health and protect against several diseases (Rodriguez et al. 2012). Added health

76 benefits of specific honey types can increase market potential as has been shown with the well known Manuka honey (Leptospermum spp.) from New Zealand and Australia. Manuka honey has considerable antibacterial activity (Allen et al. 1990), and is said to assist in a diverse range of human ailments lending to the high market price it obtains. Beekeepers can learn the predominant nectar sources of their region, and often plan harvests to keep monofloral varieties separate. While there may never be an absolute monofloral type, some honeys are relatively pure due to the prodigious nectar production of a particular species and there may be little else in bloom at the time. Beekeepers will often remove any honey made in the hive prior to the bloom of a particular desirable nectar type and remove the honey at the time the flowers are done blooming to ensure its purity. Most bees produce honey from floral nectar; however honey can also be produced when bees forage on honeydew, a sugary excretion (waste product) of phloem-feeding insects. Aphids and scale insects are common honeydew producing insects that bees will benefit from and in many geographical regions large quantities of honeydew honey are produced where these insect densities are sufficiently dense. Honeydew is the main product used by bees to make honey in many regions of the world (Crozier 1981, Kunkel 1997). World- wide and from a commercial point of view it is a relatively minor honey type, however in many European countries like Germany, Switzerland, Austria, Slovenia, Greece, Turkey and others, honeydew honey is harvested in relatively high amounts, achieving very good prices (Honeydew Symposium 2008). There are reasons why honeydew honey generally does not receive much exposure best described by Kunkel (1997) who states ”beekeepers are not interested in publicizing the fact that the most costly, tasteful honey comes from the feces of unknown insects”. Kunkel (1997) estimates that 50% of all honey produced in Middle Europe is derived from honeydew. Norway spruce and the scale insect Physokermes hemicryphus is responsible for producing much of this honey. In southern Europe, along the Aegian coast of Greece and Turkey, a Margarodid Marchalina hellenica present on Aleppo pine is a major honeydew source for bees and has been well studied (Santas 1983). In this region along the Aegian coast, it has been estimated that 65% of honey produced comes from scale insects. For other areas of Europe estimates vary between 50%-100% of all honey produced coming from honeydew. In South America, Brasil produces honeydew honey, called ’melato’ or psuedomel in the Southern part of the country on host plant Mimosa bracaatinga, a plant that is cultivated in some regions (Campos et al. 2003). New Zealand commonly calls honeydew honey ’bush honey’ or ’forest honey’. The majority of this honey is produced in the Nothofagus (southern beech) forests. Honeydew honey from New Zealand is a palatable commodity. Since scale insects can provide access to carbohydrate rich phloem in plants bees are not limited to feeding only on flowering plants. Non-flowering gymnosperms can serve as ’floral’ sources for honeydew. Also of benefit, the duration of feeding by bees is not limited to a specific window of time while the tree is inflorescent. The duration of honeydew abundance annually will depend on the species of insect producing honeydew many though provide a steady carbohydrate resource for several consecutive months within a given year (Hodgson et al. 2007). When bees feed from floral resource both pollen and nectar are usually available. With honeydew resource foraging pollen is not available and must be collected elsewhere in order to feed young developing bees within the hive. In this aspect honeydew

77 is less convenient to bees and beekeepers. Honeydew honey and forest honey is commonly a rich dark robust flavored honey, a preferred honey within many continents. It has recently been discovered that honeydew honey had exceptional antibacterial activity against drug resistant bacteria, surpassing that of even Manuka honey (Majtan et al. 2010). This antibacterial activity of honeydew honey will strengthen the marketing potential locally and globally.

5.5 Case study example: Veracruz Mexico

As an example of how these factors are relevant, I present a case study for the introduc- tion of beekeeping to Chiconquiaco, Mexico situated within tropical montane cloud forests. Montane cloud forests once covered the majority of Chiconquiaco, but this habitat has been reduced in extent because of land conversion to dairy cattle grazing, agriculture fields, and rural establishments. Montane cloud forest is a rare type of evergreen mountain forest found in tropical regions at elevations between 1500 and 3000 meters above sea level. Montane cloud forests provide important ecological services, the most notable being the potential for water capture. The montane cloud forests of northeastern Mexico have a high concentration of endemism, and are increasingly vulnerable to climate change, deforestation, and habitat fragmentation (Markham 1998; Bubb et al. 2004). Ninety percent of the original montane cloud forest in Mexico has been lost (Ramirez-Marcial et al. 2001). Some initiatives in the region have arisen to help curtail this problem, such as benefits from national initiatives that promote the importance of the ecological services that forests provide (SEMARNAT 2003). Beekeeping has not yet been formally implemented as a sustainable development tool in the tropical montane forest regions of Mexico, by generating income among the local com- munity and indigenous groups beekeeping can promote conservation of natural resources and economic development while involving the participation of local people who make the tropi- cal montane cloud forest their home. The low volcanic ranges of the Sierra de Chiconquiaco that interrupt the coastal plain of Veracruz state hold promise for the development of hon- eydew beekeeping. A honeydew-producing scale insect, Stigmacoccus garmilleri Hempel, is associated with oak trees (Quercus spp.) in highland forests of Veracruz, Mexico (Hodgson et al. 2007) (Figure 1). Ants, the usual consumers of scale insect honeydew are occasionally observed. Birds (Gamper and Koptur 2010), wasps (Vespula spp.), honeybees (Apis mel- lifera), mites (Arcari), and flies (Diptera) are more commonly found foraging on honeydew (Hodgson et al. 2007). The specific study site is located within the township of Chiconquiaco at roughly 2000m elevation. The current municipal area of Chiconquiaco is considered rural, with approxi- mately 2900 inhabitants and of these inhabitants 26% are indigenous. Dairy cattle grazing is the most dominant land-use practice in the remaining montane cloud forest of Chicon- quiaco (INEGI 2005). Thus, tropical montane forests within the region are discontinuously distributed. Remaining forests commonly form an abrupt boundary with adjacent grazed grassland, small agricultural fields, and clearings for houses and community structures. Forest fragments have low edge to interior area ratios and are comprised of a canopy of hybridizing assemblages of Quercus spp. (predominantly Q. laurina, Q. affinis, Q. crassi-

78 folia and Q. germana). All of the oaks are apparently capable of hosting the scale insect S. garmilleri. Pastures consist of small clumps and isolated individuals. Oak flowers are wind pollinated and produce no nectar reward for pollinators. The honeydew resource in these forests is extremely abundant. Excess honeydew com- monly drops to the forest floor or onto the surface of oak trees and is consumed by sooty mold (Gamper and Koptur 2010). Managed colonies of honeybees are currently not kept in these regions and could possibly thrive on this forage. It is currently not known as to how many hives this forest could support, or at what point, if any, would there be competition with migratory or endemic birds in this region. Thus for the initiation of any beekeeping project here, some consideration must be given to understanding these community level resource interactions prior to large-scale project implementation. To be of use to bees and beekeepers, the honeydew source needs to produce large amounts at any given time and within a given locality. According to Kunkel (1997) these sources require the following characteristics: the elimination of honeydew by individual insects should be high per unit time, the insects should be abundant in any given locality, and the population needs to be accessible to bees and beekeepers. The forests of Chiconquiaco satisfy these requirements (Gamper et al. 2011). Honeydew concentration measurements taken show that drops can be replaced within 30 minute time periods and individual honeydew drop volumes exceed those of other phloem feeding insects (Gamper, unpublished data). Insects are abundant in all forest patches and trees in isolated pasture areas for at least 10 km2 and populations are accessible by both paved roads and well maintained un-paved roads. When practiced on a small scale, beekeeping is not particularly labor intensive and it is possible for hives to be tended outside of regular working time. Beekeepers in the community could carry on with usual agricultural activities such as growing corn and tending dairy cows. If beekeeping were to prove profitable for the community, more hives could be tended and a shift from dairy farming and agriculture to more forest friendly activities could be achieved. It is not expected that these agricultural activities would be abandoned; however, pressure to create more agricultural fields and cattle pasture would diminish. At times when honeydew production is minimal (roughly August-October) beekeepers could explore other areas and experiment with honey production. The rich diverse Mexican flora can provide honeys with a variety of interesting flavors in addition to interesting biological properties (Prevot-Julliard et al. 2011). Identification of potential for making varieties of monofloral honey within the local area would expand the economic potential of beekeeping in Chiconquiaco. In addition beekeepers could move hives to nearby coffee, mango and avocado plantations (less than 10 km away). In return from increased yields from pollination, beekeepers could earn additional cash income from agricultural land owners. Beekeepers are typically paid for each hive they provide for pollination of fruiting crops. In addition to obtaining a profit from the sale of honey, small local landowners may also be able to obtain benefits from national initiatives that promote the importance of the ecological services that forests provide (SEMARNAT 2003). Beekeeping demands less financial investment when hives are constructed from locally available material. The management of honeybees by most beekeepers today in Mexico is done with Langstroth hives, in which construction requires more precise wood dimensions than traditional beekeeping in logs, or the Kenyan top-bar method of beekeeping. It may

79 be necessary to find external financial help to purchase the first set of woodenware. These initial boxes can be used as standards for construction of additional hive bodies or hive boxes may also be obtained through a local beekeeping cooperative, ’ejido’, within the state of Veracruz, Mexico. Initial bees can be obtained by capturing swarms in baited containers in trees, or colonies could be purchased from neighboring coffee plantations with existing beekeepers. Capturing swarms is a no-cost method to starting kept honeybee colonies. There may be advantages however in purchasing bees from Mexican apiculturalists, since the genetic stock used may yield bees that yield greater amounts of honey or are easier to manage (gentle/swarm less frequently). Beekeeping in Mexico is acknowledged as an activity of high economic, social and eco- logical importance. Mexico is the sixth foremost honey producer in the world and ranks number three for honey exports. (Guzman-Novoa 2007) In 2007 honey exports generated 69 million US dollars for the country (SIAP 2009). Beekeeping has a direct benefit on approx- imately 40000 beekeepers and their families, as well as an indirect benefit on about 400 000 people involved in apiculture related activities, such as manufacturers and suppliers of bee- keeping equipment, honey packers, and people who sell other bee products (Correa-Benitez 2004) Bees in Mexico not only help in maintaining the equilibrium of many ecosystems, by pollinating wild plants, the effect of this bee service on Mexico’s agricultural crops has an estimated value that exceeds two billion US dollars a year (Guzman-Novoa 2007). Despite its importance, the Mexican beekeeping industry is currently affected by a variety of problems, including the negative effects of Africanized bees, which are considered as one of the most damaging factors to the productivity of the industry (Guzman-Novoa and Espinosa Montano 2011). It is believed that the first swarms of Africanized bees entered Mexico through Chiapas by the end of 1986, 29 years after originating and migrating in Brazil (Moffett et al. 1987) After their arrival, Africanized bees spread during several years throughout the southeast part of the country. In 1987 they had already been found in the three states of the Yucatan peninsula, in addition to the states of Oaxaca, Tabasco and southern Veracruz (Guzman-Novoa and Espinosa Montano 2011). Africanized bees have been replacing their European counterparts as they spread throughout Mexico (Taylor et al. 1991). Currently they are established in more than 95% of the country’s beekeeping regions (Guzman-Novoa and Espinosa Montano 2011). From the biological point of view, the Africanized honey bees (descendants of Apis mellifera scutellata Lepeletier) are very successful insects, because they have been able to colonize and prevail in more than many countries. However, what Latin American beekeepers are more concerned about, is not to know if these bees are biologically successful, but whether or not they are better than races of European honey bees to maintain a lucrative beekeeping practice with them (Guzman-Novoa and Espinosa Montano 2011). African bees are well adapted to a habitat where swarming (colony division), and ab- sconding (complete colony moves and establishes elsewhere) are events that occur very frequently. They will nest in open sites, favored by the dry and hot environment of the region they originated from. Therefore, their swarms, even the smallest, have high proba- bilities of survival. Some African honey bee colonies are very defensive, which is considered to have resulted from adaptation and survival for thousands of years in an environment with high levels of predation. The main management changes implemented since the arrival of Africanized bees have

80 been of apiary management. Most of the apiaries need to be located or relocated 200m from any human housing establishments or farms, animal pens, and other locations with domestic or captive animals (Guzman-Novoa and Espinosa Montano 2011). This can be a challenge in rural areas with considerable housing settlement and livestock influence. The number of hives per apiary needs to be reduced in addition to the individual placement of hives to prevent the possibility that aggressive colonies alter the behavior of the rest of the colonies in an apiary (Guzman-Novoa and Espinosa Montano 2011). Both of these practices involve potential loss of profit and additional effort. The relocation of apiaries has been difficult to achieve in locations where few secondary roads exist (Guzman-Novoa and Page 1994).

When beekeeping is practiced with European races of bees, beekeepers can wear light clothes and a simple veil. Managing Africanized bees wearing such clothes and protective equipment could be fatal (Guzman-Novoa and Espinosa Montano 2011). Between 1988 - 2009 there was an average of 21 annual deaths related to toxic reactions caused by multiple stings from Africanized honey bees (Becerril-Angeles 2011). In most of these cases, careless behavior about honey-bee swarms, hives and colonies was the cause mostly related to deaths (Becerril-Angeles 2011) Education on safety issues surrounding Africanized bees should be made a priority in rural areas with managed colonies of Africanized bees. Beekeepers will need to invest in quality protection equipment, such as thick coveralls, boots, gloves, and better structured veils that do not touch the beekeeper’s face to reduce any harm from defensive colonies and accessive stings. Larger smokers have been built to produce more smoke for longer time, to pacify the bees (Guzman-Novoa and Page 1994). These changes imply higher initial start up costs for beekeeping implementation.

Another important management practice to control the negative effects of Africanized bees, involves requeening stock from a selected stock of European origin (Winston 1992). Before the arrival of Africanized bees, it was not common that Mexican beekeepers re- queened their colonies, but after their arrival, almost all beekeepers started to at least change the queens of their most aggressive colonies (Guzman-Novoa and Page 1994). The additional cost of queen replacement needs to be factored in on a continual basis throughout the year.

There are additional ways in which the natural history traits of Africanized bees can affect management strategies and profit loss. Because Africanized bees abscond in times of dearth, Mexican beekeepers have been confronted with the need to provide sugar syrup to their colonies in higher quantities than when they worked with European bees during these periods; this practice has increased their production costs (Guzman-Novoa and Page 1994) In addition, before the arrival of Africanized bees, Mexican beekeepers could make a single large honey harvest at the end of the blooming season. After the arrival of these bees, beekeepers realized that they had to harvest small amounts of honey from their colonies several times during the blossom season to prevent Africanized bees from using honey stores to transform them into brood (Guzman-Novoa and Page 1994). This practice implies more hours of work in the apiary, but is necessary when managing these bees.

81 5.6 Conclusions The unique qualities of this case study do not fit the typical beekeeping aid project. We recommend an approach that embraces both applied learning and adaptive management in consideration of implementation of rural beekeeping programs aimed both forest conser- vation and poverty alleviation. The foundation of this learning lies in the understanding of the environment and the biology of bees, which at times is complex; however the basic elements of understanding to manage colonies properly is within close grasp. Main considerations important to starting any beekeeping project as a mean of livelihood and resistance to deforestation include:

1. Landscape configuration and resources available to managed bees 2. Potential changes to the current ecosystem with the presence of increased densities of managed bees and the effects of those changes on other organisms. 3. Availability of a short migratory network of additional food resources and/or income from pollination services to agricultural crops. 4. Identifying and ensuring sufficient people with a diversity of skills are available for the multitude of tasks involved in a community project such as beekeeping. 5. Prior to the program establishment economic outlines, including start up costs, poten- tial and guidelines for how profits from the sale of honey, will be distributed between the beekeepers and the landowners must be in place. 6. Successful program establishment will focus on managing pests, pathogens and hive nutrition in addition to management considerations for Africanized bees if they will involved in the beekeeping project.

In the case study of Veracruz, Mexico, the following specific conditions and parameters are relevant to the establishment of a beekeeping initiative.

1. Ensure that the density of established hives does not cause resource competition with migratory and resident endemic birds. 2. Establish honeydew honey as a quality local product and research larger market po- tential through the identification of any high antibacterial or antioxidant capacity. 3. Develop a yearly management calendar where windows of opportunity for nearby avo- cado, mango and coffee pollination could be identified and a small migratory network could be formed. 4. Include all participants in the understanding of the economic realities of the program and include all skills and talents from within the community and look for nearby expertise when necessary (beekeeping cooperatives, expertise from local institutes) 5. Proper management techniques for managing Africanized bees are crucial for success- ful implementation of beekeeping in rural Chiconquiaco

82 CHAPTER 6

CONCLUSIONS

Tropical montane cloud forests are a rare forest type, contain high levels of endemic species and are important in the ecosystem services they provide, most notably for their potential for water capture. This research is a study of how land-use changes surrounding these forests can alter the dynamics of remaining tropical montane forest patches or fragments in Veracruz, Mexico. Forest fragments in this study are biotic representatives of formerly in- tact montane cloud forests. Understanding their persistence and interactions with existing land use will permit us to broaden our impacts by contributing to dialogues on their man- agement (Chazdon, 2003). This research also engages in broader perspectives surrounding the challenges of understanding how insect densities alter forest processes. Spatial patterns and dynamics within forests carry information in processes which operated in the past, and they can form the template on which processes will take place in the future (Law et al. 2009). Chapter two of this dissertation describes the importance of honeydew producing scale insects present in forest fragments and isolated oak trees in pasture areas for other organisms in the focal study area. It outlines the details of the reliance of many bird species on this rich carbohydrate resource. The consumption and even defense of scale insect honeydew by the bird community in forest trees differs from how it is used in isolated pasture trees. Of all the aggressive chases observed for scale insect honeydew only 9.65% occurred in forest observation trees, while 90.35% occurred in pasture trees. When forests are converted to more open agricultural habitats, the importance of isolated resources may increase, as is supported by the preferential defense and territorial patrolling of scale-insect honeydew by birds in scattered pasture trees. Unfortunately isolated trees in pasture areas are commonly felled or removed. This activity was witnessed on the ground during the duration of this research and quantified through the land cover changes estimated using satellite imagery classification (77.23% loss of scattered pasture tree habitat between 1989-2000). Conserva- tion priorities in this region should not only focus on the maintenance of forest patches but should also include preservation of these isolated trees, for they are important ecological resources. Chapter three reveals how land-use change has altered the distribution of this important species of scale insect, Stigmacoccus garmileri. Scale insect density, honeydew volume, and sugar concentration were surveyed throughout a continuous landscape that included both patches of forest and scattered pasture trees. Results obtained from this data collection

83 describe the increases in scale insect infestation of trees with forest disturbance. Trees located on the forest edge are capable of hosting scale insects on all portions of the tree, whereas the distribution of scale insects on forest interior trees is usually limited to the upper branches. With the tendency of forests to become more fragmented through the conversion of forests to pasture areas, the quantity of forest edge and trees exposed to conditions that lead to greater densities of scale insects increases. The effect of these increased scale insect densities on the host tree physiology is critical to understanding the forest dynamics in this region. Chapter four investigated the phys- iological effects of scale insects on oak tree growth using a dendrochronological (tree-ring study) approach. Tree cores were collected from the sample trees with varying: densities of scale insects, distance from forest edge, and tree size classes. This sampling strategy allowed for the interpretation of how these various factors affect oak tree growth. Despite estimations of little impact of scale insects (with similar amounts of tree phloem removal during feeding) on the growth of host trees in a New Zealand beech forest (Dungan et al. 2007; plant physiology modelling approach), both large and small trees in this study with low densities of scale insects grew more than similar size classes that had dense infestations of scale insects. Small trees on the forest edge exhibited very little annual growth. These findings are concerning for the future of these forest remnants. Reforestation procedures in Mexico often concentrate on planting seedlings at the edge of forest patches. In the case of the montane oak forest of Chiconquiaco, Mexico the success of this procedure will be complicated by increased scale insect densities on existing trees at the forest edge. The proximity to infected trees has been shown to increase the chances that scale insects will migrate down the canopy and feed on seedlings (Wada et al. 2000). In addition, greater densities of scale insects have been found to be detrimental to seedling establishment (Mendel et al. 1997). Any further information on the growth of edge trees with populations of scale insects will continue to aid in our understanding of how changes to biotic interactions (such as these insect-plant interactions) from human land use change can in turn alter processes such as forest regeneration and ecological restoration efforts. Future directions of this research should also consider the interaction between climate factors and effects of scale insects on the host tree. In addition a thorough understanding of how the microclimatic moisture gradient in the varying forest habitat can influence patterns in tree growth would promote understanding of the mechanisms that might influence larger scale changes to tropical forest through global climate change. In the reality of substantial forest loss and difficulties of forest recovery in this region, the research in chapter five challenges those involved in bettering rural livelihoods to rethink the approach of implementing conservation strategies that promote non-timber forest products particularly the production of honey. Specifically this study demonstrates the importance of incorporating local knowledge into productive forest conservation strategies that involve rural beekeeping. Scale insect honeydew is discussed as a case study for producing honey- dew honey with the potential in development of a sucessful rural beekeeping program in Chiconquiaco, Mexico. Sixty percent of the annual honey produced in Greece comes from bees that feed on honeydew (Bacandritsos 2002). Honeydew honey is one of New Zealand’s premium export honeys and the most renowned honeydew honeys are Black Forest honeys from Germany. As a non-timber forest product, honeydew honeys can provide a source of

84 income for hive owners or for persons renting out or safeguarding locations with access to honeydew. In addition to obtaining a profit from the sale of honey, small local landowners may also be able to obtain benefits from national initiatives that promote the importance of the ecological services that montane forests provide (SEMARNAT 2003). If scale insects are having a negative impact on the growth and regeneration of host oaks, fragmentation may be viewed as more of a ’lock-in’ or hysteresis (Walker et al. 2004), an irreversible phenomena that may lead to continued degradation of these forest fragments even with the cessation of human uses. Further intellectual merit of this research rests in its alternative view how certain species, such as scale insect Stigmacoccus garmilleri, may promote diversity in forest fragments, but at the potential expense of modifying the conditions for its own persistence. Island biogeography and metapopulation theory typically do not consider engineering effects and their influence on diversity in fragmented systems (Heaney, 2007). Furthermore, disturbance and human impacts are often the direct causal agent implicated in non equilibrium shifts to new forest states (Vale, 1982; Gunderson 2000). This dissertation documents how intrinsic dynamics set in motion after fragmentation may continue to redirect system structure and function.

85 APPENDIX A

MAPS ILLUSTRATING THE DISTRIBUTION OF SCALE INSECT, STIGMACOCCUS GARMILLERI IN TROPICAL MONTANE FORESTS OF CHICONQUIACO, MEXICO

A.1 Methods A.1.1 GPS data collection Trimble Nomad GPS units were used to record tree locations in six study plots within forest area BB, and two study plots within forest area PP A.1 from July-August 2007. At each tree location (total of 142 mapped trees) the following descriptive attribute information was collected: scale insect density at the bottom, middle and top portions of the tree, diameter at breast height of each tree (DBH) and crown size (overarching size of tree branches in canopy). Scale insect density and crown size information was collected using a categorical estimation from 0-5 with 5 being a high density of insects or large crown size. GPS points were differentially corrected and were uploaded to ArcMap (ESRI 2011) with the accompanying attribute information using Pathfinder Office software (Trimble Navigation Limited) resulting in less than 3 meters of error.

A.1.2 Geovisualization Density maps in ArcMap were created to visualize the scale insect concentration at various tree heights (low, middle, and top) in relation to the position of the tree in the forest (distance from the forest edge). The series of maps in the appendix help illustrate how forest interior trees have concentrations of insects on upper branches, whereas trees located on the forest edge have dense colonies of scale insects on lower, middle and upper portions of the tree. The relationship between forest position and scale insect densisty is not as pronounced in forest area PP. This may be due to the open and scattered park like distribution of trees in this forest area, which creates edge conditions in some forest interior trees due to the large spacing between trees.

86 Forest Site BB (six study plots)

Forest Site PP (two study plots)

Figure A.1: Distribution and relative position of the individal forest plots where scale insect distribution mapping on individual trees was conducted in 2007 Chiconquiaco, Veracruz, Mexico

87 BB Plot 1

CROWN SIZE TREE DIAMETER

Figure A.2: Trees mapped within individual forest plots (fores site BB; plot 1 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Attribute data such as tree crown size and tree trunk diameter at breast height (DBH) also aided in conceptualizing forest structure.

88 BB Plot 2

CROWN SIZE TREE DIAMETER

Figure A.3: Trees mapped within individual forest plots (fores site BB; plot 2 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Attribute data such as tree crown size and tree trunk diameter at breast height (DBH) also aided in conceptualizing forest structure.

89 BB Plot 1

Bottom Middle Top

Figure A.4: Trees mapped within individual forest plots (fores site BB; plot 1 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Data collected at each individual tree on scale insect density at varying tree heights (lower 1/3, middle 1/3, and upper 1/3) were used to create scale insect density maps.

90 BB Plot 2

Bottom Middle Top

Figure A.5: Trees mapped within individual forest plots (fores site BB; plot 2 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Data collected at each individual tree on scale insect density at varying tree heights (lower 1/3, middle 1/3, and upper 1/3) were used to create scale insect density maps.

91 BB Plot 5

Bottom Middle Top

Figure A.6: Trees mapped within individual forest plots (fores site BB; plot 5 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Data collected at each individual tree on scale insect density at varying tree heights (lower 1/3, middle 1/3, and upper 1/3) were used to create scale insect density maps.

92 PP Plot 1

Bottom Middle Top

Figure A.7: Trees mapped within individual forest plots (fores site PP; plot 1 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Data collected at each individual tree on scale insect density at varying tree heights (lower 1/3, middle 1/3, and upper 1/3) were used to create scale insect density maps.

93 PP Plot 2

Bottom Middle Top

Figure A.8: Trees mapped within individual forest plots (fores site PP; plot 2 shown here) were plotted upon an aerial photograph layer to illustrate tree position relative to forest edge. Data collected at each individual tree on scale insect density at varying tree heights (lower 1/3, middle 1/3, and upper 1/3) were used to create scale insect density maps.

94 BIBLIOGRAPHY

Chapter 1

Bach CE. 1991. Direct and indirect interactions between ants (Pheidole megacephala), scales (Coccus viridis) and plants (Pluchea indica). Oecologia 87(2): 233-239.

Bouma TJ, De Vries MB, Low E, Peralta G, Tanczos C. 2005. Trade-offs related to ecosys- tem engineering: A case study on stiffness of emerging macrophytes. Ecology 86(8): 2187-2199.

Bruno JF. 2000. Facilitation of cobble beach plant communities through habitat modifica- tion by Spartina alterniflora. Ecology 81(5): 1179-1192.

Bruno JF, Stachowicz JJ, Bertness MD. 2003. Inclusion of facilitation into ecological theory. Trends in Ecology and Evolution 18(3): 119-125.

Bubb P, May I, Miles L, Sayer J. 2004. Cloud Forest Agenda. UNEP-WCMC, Cambridge, UK.

Buckley RC. 1987. Interactions involving plants, Homoptera, and ants. Annual Review of Ecology and Systematics 18: 111-135.

Chazdon RL. 2003. Tropical forest recovery: legacies of human impact and natural distur- bances. Perspectives in Plant Ecology Evolution and Systematics 6(1-2): 51-71.

Coulson RN, McFadden BA, Pulley PE, Lovelady CN, Fitzgerald JW. 1999. Heterogeneity of forest landscapes and the distribution and abundance of the southern pine beetle. Forest Ecology and Management 114(2-3): 471-485.

Crain CM, Bertness MD. 2006. Ecosystem engineering across environmental gradients: Implications for conservation and management. Bioscience 56(3): 211-218.

Ewers R. 2002. The influence of honeydew on arthropod community composition in a New Zealand beech forest. New Zealand Journal of Ecology 26(1): 23-29.

95 Folke C, Carpenter S, Walker B, Scheffer M, Elmqvist T, Gunderson L, Holling CS. 2004. Regime shifts, resilience, and biodiversity in ecosystem management. Annual Review of Ecology Evolution and Systematics 35: 557-581.

Grilli MP, Bruno M. 2007. Regional abundance of a planthopper pest: the effect of host patch area and configuration. Entomologia Experimentalis Et Applicata 122(2): 133- 143.

Hastings A, Byers JE, Crooks JA, Cuddington K, Jones CG. 2007. Ecosystem engineering in space and time. Ecology Letters 10(2): 153-164.

INEGI. 2005. Instituto Nacional de Estadistica Geografia y Informatica (http://portal. veracruz.gob.mx).

Johnson DM, Bjornstad ON, Liebhold AM. 2006. Landscape mosaic induces traveling waves of insect outbreaks. Oecologia 148(1): 51-60.

Jones CG, Lawton JH, Shachak M. 1994. Organisms as ecosystem engineers. Oikos 69(3): 373-386.

Jones CG, Lawton JH, Shachak M. 1997. Positive and negative effects of organisms as physical ecosystem engineers. Ecology 78(7): 1946-1957.

Laurance WF. 2004. Forest-climate interactions in fragmented tropical landscapes. Philo- sophical Transactions of the Royal Society of London Series B-Biological Sciences 359(1443): 345-352.

Laurance WF. 2002. Hyperdynamism in fragmented habitats. Journal of Vegetation Science 13(4): 595-602.

Lawton JH. 1994. What do species do in ecosystems? Oikos 71(3): 367-374.

Malanson GP, Wang Q, Kupfer JA. 2007. Ecological processes and spatial patterns before, during and after simulated deforestation. Ecological Modelling 202(3-4): 397-409.

Markham A. 1998. Potential impacts of climate change on tropical forest ecosystems. Climatic Change 39(2-3): 141-143.

Moreau G, Eveleigh ES, Lucarotti CJ, Quiring DT. 2006. Ecosystem alteration modifies the relative strengths of bottom-up and top-down forces in a herbivore population. Journal of Animal Ecology 75(4): 853-861.

Pool CA. 2006. Current research on the Gulf Coast of Mexico. Journal of Archaeological Research 14(3): 189-241.

96 Ramirez-Marcial N, Gonzalez-Espinosa M, Williams-Linera G. 2001. Anthropogenic distur- bance and tree diversity in Montane Rain Forests in Chiapas, Mexico. Forest Ecology and Management 154: 311-326.

Reichman OJ, Seabloom EW. 2002. The role of pocket gophers as subterranean ecosystem engineers. Trends in Ecology and Evolution 17(1): 44-49.

Stachowicz JJ. 2001. Mutualism, facilitation, and the structure of ecological communities. Bioscience 51(3): 235-246.

Stallins JA. 2006. Geomorphology and ecology: Unifying themes for complex systems in biogeomorphology. Geomorphology 77: 207-216. van Nes EH, Scheffer M. 2004. Large species shifts triggered by small forces. American Naturalist 164(2): 255-266.

Williams-Linera G, Devall MS, Alvarez-Aquino C. 2000. A relict population of Fagus gran- difolia var. mexicana at the Acatlan Volcano, Mexico: structure, litterfall, phenology and dendroecology. Journal of Biogeography 27(6): 1297-1309.

Wilson JB, Agnew ADQ. 1992. Positive-feedback switches in plant communities. Adv. Ecol. Res. 23: 263- 336.

Wright JP, Jones CG. 2006. The concept of organisms as ecosystem engineers ten years on: progress, limitations, and challenges. BioScience 56(3):203-209.

Chapter 2

Bach CE. 1991. Direct and indirect interactions between ants (Pheidole megacephala), scales (Coccus viridis) and plants (Pluchea indica). Oecologia 87:233-239.

Beggs J. 2001. The ecological consequences of social wasps (Vespula spp.) invading an ecosystem that has an abundant carbohydrate resource. Biological Conservation 99:17- 28.

Buckley R. 1987. Ant-plant-homopteran interactions. Advances in Ecological Research 16:53-85.

Cayuela L, Golicher DJ, Benayas JMR, Gonzalez-Espinosa M, Ramirez-Marcial N. 2006. Fragmentation, disturbance and tree diversity conservation in tropical montane forests. Journal of Applied Ecology 43:1172-1181.

Clements JF. 2000. Birds of the world: a checklist. (Fifth edition). Ibis Publishing Com- pany, Vista. 867 pp.

97 Daily GC, Ehrlich PR, Sanchez-Azofeifa. 2001. Countryside biogeography: utilization of human-dominated habitats by the avifauna of southern Costa Rica. Ecological Appli- cations 11:1-13.

Edwards EP. 1982. Hummingbirds feeding on an excretion produced by scale insects. Con- dor 84:122.

Gaze PD, Clout MN. 1983. Honeydew and its importance to birds in beech forests of South Island, New Zealand. New Zealand Journal of Ecology 6:33-37.

Gibbons P, Linednmayer DB, Fischer J, Manning AD, Weinberg A, Seddon J, Ryan P, Barrett G. 2008. The future of scattered trees in agricultural landscapes. Conservation Biology 22:1309-1319.

Graham C, Martinez-Leyva JE, Paredes LC. 2002. Use of fruiting trees by birds in contin- uous forest and riparian forest remnants in Los Tuxtlas, Veracruz, Mexico. Biotropica 34:589-597.

Greenberg R, Caballero CM, Bichier P. 1993. Defense of homopteran honeydew by birds in the Mexican highlands and other warm temperate forests. Oikos 68:519-524.

Greenberg R, Reitsma R, Cruz Angon A. 1996. Interspecific aggression by Yellow Warblers in a sun coffee plantation. Condor 98:640-642.

Hodgson C, Gamper H, Bogo A, Watson G. 2007. A taxonomic review of the Margarodoid genus Stigmacoccus Hempel (Hemiptera: Sternorrhyncha: Coccoidea: Stigmacoccidae), with some details on their biology. Zootaxa 1507:1-55.

Hughes JB, Daily GC, Ehrlich PR. 2002. Conservation of tropical forest birds in countryside habitats. Ecology Letters 5:121-129.

Jiron, LF, Salas S. 1975. Simbiosis entre cochinillas de cola (Coccoidea: Margarodidae) y otros insectos. Brenesia 5:67-71.

Johnson MD, Sherry TW. 2001. Effects of food availability on the distribution of migratory warblers among habitats in Jamaica. Journal of Animal Ecology 70:546-560.

Koster F, Stoewesand H. 1973. Schildluse als Honigtaulieferanten fr Kolibris und Insekten. Bonner Zoologische Beitraege 24:15-23.

Latta SC, Faaborg J. 2002. Demographic and population responses of Cape May Warblers wintering in multiple habitats. Ecology 83:2502-2515.

Latta SC, Gamper HA, Tietz JR. 2001. Revising the convergence hypothesis of avian use of honeydew: evidence from Dominican subtropical dry forest. Oikos 93:250-259.

98 Levey, DJ. 1988. Spatial and temporal variation in Costa Rican fruit and fruit-eating bird abundance. Ecological Monographs 58:251-269.

Loyn RH, Runnals RG, Forward GY, Tyers J. 1983. Territorial bell miners and other birds affecting populations of insect prey. Science 221:1411-1413.

Manning AD, Fischer J, Lindenmayer DB. 2006. Scattered trees are keystone structures - implications for conservation. Biological Conservation 132:311-321.

Moller H, Tilley JAV. 1989. Beech honeydew: seasonal variation and use by wasps, honey- bees, and other insects. New Zealand Journal of Zoology 16:289-302.

Murphy DJ, Kelly D. 2001. Scarce or distracted? Bellbird (Anthornis melanura) foraging and diet in an area of inadequate mistletoe pollination. New Zealand Journal of Ecology 25:69-81.

Murphy DJ, Kelly D. 2003. Seasonal variation in the honeydew, invertebrate, fruit and nectar resource for bellbirds in a New Zealand mountain beech forest. New Zealand Journal of Ecology 27:11-23.

Orians GH, Willson MF. 1964. Interspecific territories of birds. Ecology 45:736-745.

Paton DC. 1980. The importance of manna, honeydew, and lerp in the diets of honeyeaters. Emu 80:213-226.

Petit DR, Lynch JF, Hutto RL, Blake JG, Waide RB. 1995. Habitat use and conservation of migratory landbirds wintering in the Neotropics. Pp. 145-200 in Martin, T. E. and Finch, D. (eds.). Ecology and Management of Neotropical Migratory Birds. Oxford University Press, New York.

Reichholf H, Reichholf J. 1973. Honigtau der Bracaatinga-Schildlaus als Winternahrung fr Kolibris (Trochilidae) in Sd-Brasilien. Bonner Zoologische Beitraege 24:7-14.

Robbins CS, Dowell BA, Dawson DK, Colon JA, Estrada R, Sutton A, Sutton R, Weyer D. 1992. Comparison of Neotropical migrant landbird populations wintering in tropical forest, isolated forest fragments, and agricultural habitats. Pp. 207-220 in Hagan III, J. M. and Johnston, D.W. (eds.) Ecology and Conservation of Neotropical Migrant Landbirds. Smithsonian Institution Press, Washington, D.C. 609 pp.

Watson DM. 2003. Long-term consequences of habitat fragmentation-highland birds in Oaxaca, Mexico. Biological Conservation 111:283-303.

Way MJ. 1963. Mutualism between ants and honeydew-producing Homoptera. Annual Review of Entomology 8:307-344.

99 Williams JR, and Williams DJ. 1980. Excretory behaviour in soft scales (Hemiptera: Coc- cidae). Bulletin of Entomological Research 70:253-257.

Williams-Linera G, Devall MS, Alvarez-Aquinoa C. 2000. A relict population of Fagus grandifolia var. emph{mexicana at the Acatlan Volcano, Mexico: structure, litterfall, phenology and dendroecology. Journal of Biogeography 27:1297-1309.

Woinarski JCZ. 1984. Small birds, lerp-feeding and the problem of honeyeaters. Emu 84:137-141.

Chapter 3

Bubb P, May I, Miles L, Sayer J. 2004. Cloud Forest Agenda. UNEP-WCMC - World Conservation Monitoring Centre.

Crozier LR. 1981. Beech honeydew: forest produce. New Zealand Journal of Forestry 26: 200-209.

Didham RK. 1993. The influence of honeydew on arthropods associated with beech trees in New Zealand. New Zealand Natural Sciences 20: 47-53.

Didham RK, Fagan LL. 2004. Forest Canopies. In: Burley J, Evans J, Youngquist J, Editors. Encyclopedia of Forest Sciences. pp. 68-80. Academic Press.

Dungan RJ, Kelly D. 2003. Effect of host-tree and environmental variables on honeydew production by scale insects (Ultracoelostoma sp.) in a high elevation Nothofagus solandri forest. New Zealand Journal of Ecology 27: 169-177.

Dungan RJ, Kelly D, Turnbull M. 2007. Separating host-tree and environmental determi- nants of honeydew production by Ultracoelostoma scale insects in a Nothofagus forest. Ecological Entomology 32: 338-348.

Ewers RM, Didham RK. 2006. Confounding factors in the detection of species responses to habitat fragmentation. Biological Reviews 81: 117-142.

Foldi I. 1995. Margarodidae du Mexique (Hemiptera: Coccoidea). Annales de la Socit Entomologique de France (Nouvelle Srie) 31: 165-178.

Gamper H, Koptur S. 2010. Honeydew foraging by birds in tropical montane forests and pastures of Mexico. Journal of Tropical Ecology 26: 335-341.

Gaze PD, Clout MN. 1983. Honeydew and its importance to birds in beech forest of the South Island, New Zealand. New Zealand Journal of Ecology 6: 33-37.

100 Geraud-Pouey F, Chirinos DT, Romay G. 2001. Physical effect of exfoliation of guava tree bark on Capulinia sp. near to jaboticabae von Ihering (Hemiptera: Eriococcidae). Enotomotropica 16: 21-27.

Gibbons P, Lindenmayer DB, Fisher J, Manning AD, Weinberg A, Seddon J, Ryan P, Barrett G. 2008. The future of scattered trees in agricultural landscapes. Conservation Biology 22: 1309-1319.

Grant WD, Beggs JR. 1989. Carbohydrate analysis of beech honeydew. New Zealand Journal of Zoology 16: 283-288.

Greenberg R, Caballero CM, Bichier P. 1993. Defense of homopteran honeydew by birds in the Mexican highlands and other warm temperate forests. Oikos 68: 519-524.

Gullan PJ, Kosztarab M. 1997. Adaptations in scale insects. Annual Review of Entomology 42: 23-50.

Hodgson C, Gamper H, Bogo A, Watson G. 2007. A taxonomic review of the Margarodoid genus Stigmacoccus Hempel (Hemiptera: Sternorrhyncha: Coccoidea: Stigmacoccidae), with some details on their biology. Zootaxa 1507: 1-55.

Inouye DW, Favre ND, Lanum JA, Levine DM, Meyers JB, Roberts FC, Tsao FC, Wang Y. 1980. The effects of non-sugar nectar constituents on nectar energy content. Ecology 61: 992-996.

James A, Dungan R, Plank M, Ito R. 2007. A dynamical model of honeydew droplet pro- duction by sooty-beech scale insects (Ultracoelostoma spp.) in New Zealand Nothofagus forest. Ecological Modeling 209: 323-332.

Kelly D. 1990. Honeydew density in mixed Nothofagus forest, Westland, New Zealand. New Zealand Journal of Botany 28: 53-58.

Kelly D, Stirling DJ, Hunt GR, Newell CL, Jarvis CE. 1992. Honeydew standing crop and production over 24 hours in Nothofagus solandri forest in Canterbury. New Zealand Journal of Ecology 16: 69-75.

Latta SC, Gamper HA, Tietz JR. 2001. Revising the convergence hypothesis of avian use of honeydew: evidence from Dominican subtropical dry forest. Oikos 93: 250-259.

Laurance WF. 2002. Hyperdynamism in fragmented habitats. Journal of Vegetation Science 13: 595-602.

Malanson GP, Wang Q, Kupfer JA. 2007. Ecological processes and spatial patterns before, during and after simulated deforestation. Ecological Modeling 202: 397-409.

101 Manning AD, Gibbons P, Lindenmayer DB. 2009. Scattered trees: a complementary strat- egy for facilitating adaptive responses to climate change in modified landscapes? Journal of Applied Ecology 46: 915-919.

Markham A. 1998. Potential impacts of climate change on tropical forest ecosystems. Climatic Change 39: 141-143.

Moller H, Tilley JAV. 1989. Beech honeydew: seasonal variation and use by wasps, honey- bees and other insects. New Zealand Journal of Zoology 16: 289-302.

Morales CF. 1991. Margarodidae (Insecta: Hemiptera). In: Crosby T, Editor. Fauna of New Zealand Ko te Aitanga Pepeke o Aotearoa. FNZ 21. Available online: www. landcareresearch.co.nz/research/biosystematics/invertebrates/faunaofnz/

Murphy DJ, Kelly D. 2003. Seasonal variation in the honeydew, invertebrate, fruit and nectar resource for bellbirds in a New Zealand mountain beech forest. New Zealand Journal of Ecology 27: 11-23.

Nicolai V. 1986. The bark of trees: thermal properties, microclimate and fauna. Oecologia 69: 148-160.

Ramirez-Marcial N, Gonzalez-Espinosa M, Williams-Linera G. 2001. Anthropogenic distur- bance and tree diversity in Montane Rain Forests in Chiapas, Mexico. Forest Ecology and Management 154: 311-326.

Wackers F. 2000. Do oligosaccharides reduce the suitability of honeydew for predators and parasitoids? Oikos 90: 197-201.

Wardhaugh CW, Blakely TJ, Greig H, Morris PD, Barnden A, Rickard S, Atkinson B, Fagan LL, Ewers RM, Didham RK. 2006. Vertical stratification in the spatial distribution of the beech scale insect (Ultracoelostoma assimile) in Nothofagus tree canopies in New Zealand. Ecological Entomology 31: 185-195.

Wardhaugh CW, Didham RK. 2006. Preliminary evidence suggests that beech scale insect honeydew has a negative effect on terrestrial litter decomposition rates in Nothofagus forests of New Zealand. New Zealand Journal of Ecology 30: 279-284.

Williams JR, Williams DJ. 1980. Excretory behaviour in soft scales (Hemiptera: Coc- coidae). Bulletin of Entomological Research 70: 253-257.

Williams-Linera G, Devall MS, Alvarez-Aquino C. 2000. A relict population of Fagus gran- difolia var. mexicana at the Acatlan Volcano, Mexico: structure, litterfall, phenology and dendroecology. Journal of Biogeography 27: 1297-1309.

102 Chapter 4

Ayres MP, Lombardero MJ. 2000. Assessing the consequences of climate change for forest herbivore and pathogens. Science of the Total Environment 262: 263-286.

Baddeley A, Turner R. 2005. Spatstat: an R package for analyzing spatial point patterns. Journal of Statistical Software 12(6): 1-42.

Biondi F, Qeadan F. 2008. A theory-driven approach to tree-ring standardization: Defining the biological trend from expected basal area increment. Tree-Ring Research 64(2): 81- 96.

Bunn A. 2008. A dendrochronology program library in R (dplR). Dendrochronologia 26:115- 124.

Clark JS. 1991. Disturbance and tree life history on the shifting mosaic landscape. Ecology 72: 1102-1118.

Coley PD, Bryant JP, Chapin FS. 1985. Resource availability and plant anti-herbivore defense. Science 230: 895-899

Coster C. 1927. Zur Anatomie und Physiologie der Zuwachszonen und Jahresbildung in den Tropen. Annales du Jardin Botanique de Buitenzorg 37: 49-160.

Chowdhury MQ, Schmitz N, Verheyden A, Sass-Klaassen U, Koedam N, Beeckman H. 2008. Nature and periodicity of growth rings in two Bangladeshi mangrove species. International Association of Wood Anatomists 29: 265276

Dungan RJ, Turnbull MH, Kelly D. 2007. The carbon costs for host trees of a phloem- feeding herbivore. Journal of Ecology 95: 603-613.

Diggle PJ. 1983. Statistical Analysis of Spatial Point Patterns. Academic Press, London.

Doak DF. 1992. Lifetime impacts of herbivory for a perennial plant. Ecology 73: 2086-2099.

Edwards EP. 1982. Hummingbirds feeding on an excretion produced by scale insects. Con- dor 84: 122.

ESRI 2011. ArcGIS Desktop: Release 10. Redlands, CA: Environmental Systems Research Institute.

Fritts H. 2001. Tree Rings and Climate. Blackburn Press, Caldwell, N.J. 567pp.

103 Gamper HA, Koptur S. 2010. Examining the use of honeydew by birds in tropical montane forests and pastures of Veracruz, Mexico. Journal of Tropical Ecology. 26: 335-341.

Gamper H, Koptur S, Garca-Franco J, Plata Stapper P. 2011. Alteration of forest structure modifies the distribution of scale insect, Stigmacoccus garmilleri, in Mexican tropical montane cloud forests. Journal of Insect Science, Vol 11: Article 120.

Getis A. 1984. Interaction Modeling Using Second-order Analysis. Environment and Plan- ning A 16: 173-183.

Greenberg R, Caballero CM, Bichier P. 1993. Defense of homopteran honeydew by birds in the Mexican highlands and other warm temperate forests. Oikos 68: 519-524.

Grissino-Mayer HD. 2001. Evaluating crossdating accuracy: A manual and tutorial for the computer program COFECHA. Tree-Ring Research 57(2): 205-221.

Gullan PJ, Kosztarab M. 1997. Adaptations in scale insects. Annual Review of Entomology 42: 23-50.

Hodgson C, Gamper H, Bogo A, Watson G. 2007. A taxonomic review of the Margarodoid genus Stigmacoccus Hempel (Hemiptera: Sternorrhyncha: Coccoidea: Stigmacoccidae), with some details on their biology. Zootaxa 1507: 1-55.

Holmes RL. 1983. Computer-assisted quality control in tree-ring dating and measurement. Tree-Ring Bulletin 43: 6978.

Illian J, Penttinen A, Stoyan H, Stoyan D. 2008. Statistical Analysis and Modelling of Spatial Point Patterns. Wiley, Chichester, UK.

Karban R. 1980. Periodical cicada nymphs impose periodical oak tree wood accumulation. Nature 287:326-327.

Karban R, Baldwin IT. 1997. Induced Responses to Herbivory. University of Chicago Press, Chicago

Kashian DM, Jackson RM, Lyons HD. 2011. Forest structure altered by mountain pine beetle outbreaks affects subsequent attack in a Wyoming lodgepole pine forest, USA. Canadian Journal of Forest Research 41 (12): 2403-2412.

Kayes LJ, Tinker DB. 2011. Forest structure and regeneration following a mountain pine beetle epidemic in southeastern Wyoming. Forest Ecology and Management 263: 57-66.

Koenig WD, Liebhold AM. 2003. Regional impacts of periodical cicadas on oak radial increment. Canadian Journal of Forest Research 33: 1084-1089.

104 Law R, Illian J, Burslem DFRP, Gratzer G, Gunatilleke CVS, Gunatilleke IAUN. 2009. Eco- logical information from spatial patterns of plants: insights from point process theory. Journal of Ecology 97: 616-628.

Lieberman D, Lieberman M, Hartschorn GS, Peralta R. 1985. Growth rates and age-size relationships of tropical wet forest trees in Costa Rica. Journal of Tropical Ecology 1: 97-109.

Lopez, L. and Villalba, R. 2011. Climate Influences on the Radial Growth of Centrolo- bium microchaete, a Valuable Timber Species from the Tropical Dry Forests in Bolivia. Biotropica 43: 4149.

Morrow PA, LaMarche VC. 1978. Tree ring evidence for chronic insect suppression of productivity in subalpine eucalyptus. Science 201(4362): 1244.

Medley KE, Pobocik CM, Okey BM. 2003. Historical changes in forest cover and land ownership in a Midwestern U.S. landscape. Annals of the Association of American Geographers 93(1): 104-120.

Phipps RL. 2005. Some geometric constraints on ring-width trend. Tree-Ring Research 61(2): 73-76.

R Development Core Team. 2004. R: A language and environment for statistical computing. Vienna, Austria: R foundation for Statistica Computing Version 2.14.

Ripley BD. 1976. The second-order analysis of stationary point processes. Journal of Applied Probability 13: 255-266.

Rozendaal DMA, Zuidema PA. 2011. Dendroecology in the tropics: a review. Trees 25: 3-16.

Schongart J, Piedade MTF, Ludwigshausen S, Horna V, Worbes M. 2002. Phenology and stem-growth periodicity of tree species in Amazonian floodplain forests. Journal of Tropical Ecology 18: 581597.

Spaulding HL, Rieske LK. 2010. The aftermath of an invasion: Structure and composition of Central Appalachian hemlock forests following establishment of the hemlock woolly adelgid, Adelges tsugae. Biological Invasions 12(9): 3135-3143.

Speer JH, Clay K, Bishop G, Creech M. 2010. The effect of periodical cicadas on growth of five tree species in Midwestern deciduous Forests. The American Midland Naturalist 164: 173-186.

Speer JH. 2010. Fundamentals of Tree-Ring Research. The University of Arizona Press. 324pp.

105 Stahle DW. 1999. Useful strategies for the development of tropical tree-ring chronologies. 20(3): 249-253.

Stokes MA, Smiley TL. 1968. An Introduction to Tree-Ring Dating. University of Chicago Press, Chicago, IL, 73 pp.

Stoyan D, Penttinen A. 2000. Recent applications of point process methods in forestry statistics. 15(1): 61-78.

Teskey RO. 1997. Combined effects of elevated CO2 and air temperature on carbon assim- ilation of Pinus taeda trees. Plant, Cell and Environment 20: 373-380.

Trotter RT, Cobb NS, Whitham TG. 2002. Herbivory, plant resistance, and climate in the tree ring record: Interactions distort climatic reconstructions. Proceedings of the National Academy of Sciences 99(15): 10197-10202.

Voorhess N. 2000. Voortech Consulting. Project J2X software.

Wackers F. 2000. Do oligosaccharides reduce the suitability of honeydew for predators and parasitoids? Oikos 90: 197-201.

Williams JR, Williams DJ. 1980. Excretory behaviour in soft scales (Hemiptera: Coccidae). Bulletin of Entomological Research 70: 253-257.

Williams-Linera G, Devall MS, Alvarez-Aquino C. 2000. A relict population of Fagus gran- difolia var. mexicana at the Acatlan Volcano, Mexico: structure, litterfall, phenology and dendroecology. Journal of Biogeography 27(6): 1297-1309.

Worbes M. 1999. Annual growth rings, rainfall dependent growth patterns of tropical trees from the Caparo Forest Reserve in Venezuela. Journal of Ecology 87(3): 391-403.

Yamaguchi DK. 1991. A simple method for cross-dating increment cores from living trees. Canadian Journal of Forest Research 21: 414416.

Yang LH. 2006. Periodical cicadas use light for oviposition site selection. Proceedings of the Royal Society B: Biological Sciences 273: 2993-3000.

Yang LH, Karban R. 2009. Long-term habitat selection and chronic root herbivory: Ex- plaining the relationship between periodical cicada density and tree growth. The Amer- ican Naturalist 173: 105-112.

106 Chapter 5

Aguiree RD, Pasteur K. 1998 Women beekeepers in Calakmul Mexico. Bees for Development 46: 6-7.

Ahmad F, Joshi SR, Gurung MB, Partap U. 2002. Women and Beekeeping in Nepal. Bees and Development 62:10.

Adgaba N, Bekele W, Ejigu K. 2008. The role of women, and indigenous knowledge in Ethiopian beekeeping. Bees for Development 86: 4-5.

Aizen MA, Feinsinger P. 2003. Bees not to be? Responses of insect pollinator faunas and flower pollination to habitat fragmentation. In: Bradshaw GA, Marquet PA (eds). How landscapes change: human disturbance and ecosystem fragmentation in the Americas. Springer-Verlag, Berlin. Pp. 111-129.

Allen KL, Molan PC, Reid GM. 1990. A survey of the antibacterial activity of some New Zealand honeys. Journal of Pharmacy and Pharmacology. 43(12): 817-22.

Barthell JF, Randall JM, Thorp RW, Wenner AM. 2001 Promotion of seed set in yellow star-thistle by honey bees: Evidence of an invasive mutualism. Ecological Applications 11(6): 1870-1883.

Beekman M, Ratnieks FLW. 2001. Long-range foraging by the honey-bee, Apis mellifera L. Functional Ecology 14(4):490-496.

Becerril-Angeles M. 2011. Accidents and mortality from honey bee stings in Mexico. The Journal of Allergy and Clinical Immunology 127(2): 249.

Becker P, Moure JS, Peralta FJA. 1991. More about euglossine bees in Amazonian forest fragments. Biotropica 23: 586-591.

Bradbear N, Food and Agriculture Organization of the United Nations. 2009. Bees and their role in forest livelihoods: A guide to the services provided by bees and the sustainable harvesting, processing and marketing of their products. Rome: Food and Agriculture Organization of the United Nations.

Bradbear N, Fisher E, Jackson H. (eds). 2002. Strengthening Livelihoods: exploring the role of beekeeping in development. Bees for Development, UK.

Brown CJ. 2001. Responding to deforestation: productive conservation, the World Bank and beekeeping in Rondonia, Brazil. Professional Geographer 53(1): 106-118.

107 Brosi BJ. 2009. The complex responses of social stingless bees (Apidae: Meliponini) to tropical deforestation. Forest Ecology and Management 258(9): 1830-1837.

Bubb P, May I, Miles L, Sayer J. 2004. Cloud Forest Agenda. UNEP-WCMC, Cambridge, UK.

Burkle LA, Alarcon R. 2011. The future of plant-pollinator diversity: understanding in- teraction networks across time, space, and global-change. American Journal of Botany 98(3): 528-538.

Butz Huryn VM. 1997. Ecological impacts of introduced honey bees. Quarterly Review of Biology 72: 275-297.

Campos G, Della-Modesta RC, Silva TJP, Baptista KE, Gomides MF, Godoy RL. 2003. Classificacao do mel em floral ou mel de melato. Cincia e Tecnologia de Alimentos 23(1): 1-5.

Cane JH. 2001. Habitat fragmentation and native bees: a premature verdict? Conservation Ecology 5(1): 3.

Chacoff NP, Aizen MA. 2006. Edge effects on flower-visiting insects in grapefruit plantations bordering premontane subtropical forest. Journal of Applied Ecology 43(1): 18-27.

Charnley S, Poe MR. 2007. Community Forestry in Theory and Practice: Where Are We Now? Annual Review of Anthropology 36: 301-336.

Chemas A, Rico Gray V. 1991. Apiculture and management of associated vegetation by the Maya of Tixcacaltuyub, Yucatan, Mexico. Agroforestry Systems. 13(1): 13-25.

Crozier LR. 1981 Beech honeydew: forest produce. New Zealand Journal of Forestry 26: 200-209.

Correa-Benitez A. 2004. Historia de la apicultura en Mexico. Imagen Veterinaria 4: 4-6.

Dohzono I, Yokoyama J. 2010. Impacts of alien bees on native plant-pollinator relationships: A review with special emphasis on plant reproduction. Applied Entomology and Zoology 45(1): 37-47.

Dustmann J. 1993. Honey, quality and its control. American Bee Journal 133: 648-651.

Eden S. 2009. The work of environmental governance networks: Traceability, credibility and certification by the Forest Stewardship Council. Geoforum 40(3): 383-394.

Evans JD, Schwarz RS. 2011. Bees brought to their knees: microbes affecting honey bee health. Trends in Microbiology 19(12): 614-620.

108 Gamper HA, Koptur S. 2010. Examining the use of honeydew by birds in tropical montane forests and pastures of Veracruz, Mexico. Journal of Tropical Ecology. 26: 335-341.

Gamper H, Koptur S, Garcia-Franco J, Plata Stapper P. 2011. Alteration of forest structure modifies the distribution of scale insect, Stigmacoccus garmilleri, in Mexican tropical montane cloud forests. Journal of Insect Science Vol 11: Article 120.

Garcia-Fernandez C, Ruiz-Perez M, Wunder S. 2008. Is multiple-use forest management widely implementable in the tropics? 256(7): 1468-1476.

Garibaldi LA, Steffan-Dewenter I, Kremen C, Morales JM, Bommarco R, Cunningham SA, Carvalheiro LG, Chacoff NP. 2011. Stability of pollination services decreases with isolation from natural areas despite honey bee visits. Ecology Letters 14(10): 1062-1072.

Goulson D, Sparrow K. 2009. Evidence for competition between honeybees and bumblebees: effects on bumblebee worker size. Journal of Insect Conservation 13(2): 177-181.

Goulson D. 2003. Effects of introduced bees on native ecosystems. Annual Review of Ecology Evolution and Systematics 34: 1-26.

Gross CL, Mackay D. 1998. Honeybees reduce fitness in the pioneer shrub Melastoma affine (Melastomataceae). Biological Conservation 86(2): 169-178.

Guzman-Novoa E, Espinosa Montano LG. 2011. Colonization, impact and control of Africanized honey bees in Mexico. Veterinaria Mexico 42(2): 149-178.

Guzman-Novoa E, Goodman RD, Huang ZY, Morse RA, Reid M, Yoshida T. 2007. Bee- keeping in various parts of the world. In: Shimanuki H, Flottum K, Harman A (eds). The ABC and XYZ of Bee Culture. Medina, OH: AI Root Co, pages 83-99.

Guzman-Novoa E, Page RE. 1994. The impact of Africanized bees on Mexican beekeeping. American Bee Journal 134: 101-106.

Hall A. 1997. Sustaining Amazonia: Grassroots action for productive conservation. Manch- ester, UK: Manchester University Press.

Hansen MC, Stehman SV, Potapov PV, Loveland TR, Townshend JRG, DeFries RS, Pittman KW, Arunarwati B, Stolle F, Steininger MK, Carroll M, DiMiceli C. 2008. Humid trop- ical forest clearing from 2000 to 2005 quantified by using multitemporal and multireso- lution remotely sensed data. Proceedings of the National Academy of Sciences 105(27): 9439-9444.

Honeydew Symposium, 2008.http://www.bee-hexagon.net/files/file/fileE/IHC-Conferences/ Tzarevo/HoneydewSymposiumProgramAbstract.pdf

109 Hodgson C, Gamper H, Bogo A, Watson G. 2007. A taxonomic review of the Margarodoid genus Stigmacoccus Hempel (Hemiptera: Sternorrhyncha: Coccoidea: Stigmacoccidae), with some details on their biology. Zootaxa 1507: 1-55.

Holzschuh A, Dorman CF, Tscharntke T, Steffan-Dewenter I. 2011. Expansion of mass- flowering crops leads to transient pollinator dilution and reduced wild plant pollination. Proceedings of the Royal Society B- Biological Sciences 278(1723): 3444-3451.

Horn T. 2011. Beeconomy: What Women and Bees Can Teach Us about Local Trade and the Global Market. Lexington: University Press of Kentucky.

Illgner PM, Nel EL, Robertson MP. 1998. Beekeeping and local self-reliance in rural South- ern Africa. The Geographical Review 88(3): 349-362.

INEGI 2005. Instituto Nacional de Estadistica Geografia y Informatica http://portal. veracruz.gob.mx.

Ingram V, Njikeu J. 2011. Sweet, sticky, and sustainable social business. Ecology and Society 16(1): 37.

Johnson R. 2010. Honey Bee Colony Collapse Disorder. Congressional Research Service Report for Congress (www.crs.gov).

Jones CG, Lawton JH, Shachak M. 1994. Organisms as ecosystem engineers. Oikos 69: 373-386.

Kainer KA, DiGiano ML, Duchelle AE, Wadt LHO, Bruna E, Dain JL. 2009. Partnering for greater success: local stakeholders and research in tropical biology and conservation. Biotropica 41(5):555-562.

Kiss A. 2004. Is community-based ecotourism a good use of biodiversity conservation funds? Trends in Ecology and Evolution 19(5): 232-237.

Klooster D. 2010. Standardizing sustainable development? The Stewardship Council’s plantation policy review process as neoliberal environmental governance. Geoforum 41(1): 117-129.

Kojima Y, Taku T, Tomomi M. 2011. Infestation of Japanese native honey bees by tracheal mite and virus from non-native European honey bees in Japan. Microbial Ecology 62(4): 895-906.

Kunkel H. 1997. Scale insect honeydew as forage for honey production. Pp. 291-302 in Ben-Dov Y, Hodgson CJ (eds.) Soft scale insects: their biology, natural enemies, and control. Elsevier, Amsterdam.

110 Kremen C, Williams NM, Bugg RL, Fay JP and Thorp RW. 2004. The area requirements of an ecosystem service: crop pollination by native bee communities in California. Ecology Letters 7: 11091119.

Labougle JM, Zozaya JA. 1986. La apicultura en Mexco. Cienca y Desarollo 12: 17-36.

Lalika MCS, Machangu JS. 2008. Beekeeping for income generation and coastal forest conservation in Tanzania. BfD Journal No. 88: 3.

Levy S. 2011. The pollinator crisis: what’s best for bees. Nature 479: 164-165.

Liu H, Pemberton RW. 2009. Invasive orchid bee outperforms co-occurring native bees to promote fruit set of an invasive Solanum. Oecologia 159(3): 515-525.

Loarie SR, Duffy PB, Hamilton H, Asner GP, Field CB, Ackerly DD. 2009. The velocity of climate change. Nature 462(7276): 1052-1055.

Louveaux J, Maurizio A, Vorwohl G. 1978. Methods of melissopalynology. Bee World 59: 139-157.

Majtan J, Majtanova L, Bohova J, Majtan V. 2010. Honeydew honey as a potent antibac- terial agent in eradication of multi-drug resistant Stenotrophomonas maltophilia isolates from cncer patients. Phytotherapy Research 25(4): 584-587.

Markham A. 1998. Potential impacts of climate change on tropical forest ecosystems. Climatic Change 39(2-3): 141-143.

Maxwell JT, Knapp PA. 2012. Reconstructed tupelo-honey yield in northwest Florida inferred from Nyssa Ogeche tree-ring data: 18502009. Agriculture, Ecosystems and Environment 149: 100108.

Molan P. 1998. The limitations of the methods of identifying the floral source of honeys. Bee World 97: 59-68.

Moffett JO, Maki DL, Andere T, Fierro MM. 1987. The Africanized bee in Chiapas, Mexico. American Bee Journal 127: 517-520.

Moretto G, Mello LJ. 2000. Resistance of Africanized bees (Apis mellifera L.) as a cause of mortality of the mite Varroa jacobsoni Oud. in Brazil. 140(11): 895-897.

Moritz RFA, Hartel S, Neumann P. 2005. Global invasions of the western honeybee (Apis mellifera) and the consequences for biodiversity. Ecoscience 12(3): 289-301.

Mutersbaugh T. 2009. Brewing justice: Fair trade coffee, sustainability, and survival. Eco- nomic Geography 85(2): 241-242.

111 Porter-Bolland L. 2003. Apiculture and the mayan countryside. A study of the flowering phenology of the melliferous species and its relation to beekeeping in the La-Montana region of Campeche, Mexico. Mexican Studies 19(2): 303-330.

Paine RT. 1995. A conversation on refining the concept of keystone species. Conservation Biology 9(4): 962-964.

Persson AS, Smith HG. 2011. Bumblebee colonies produce larger foragers in complex landscapes. Basic and Applied Ecology 12 (8): 695-702.

Pinto MA, Rubink WL, Patton JC, Coulson RN, Johnston JS. 2005. Africanization in the United States: replacement of feral European honeybees (Apis mellifera L.) by an African hybrid swarm. Genetics 170(4): 1653-65.

Prevot-Julliard AC, Clavel J, Teillac-Deschamps P, Julliard R. 2011. The need for flexibility in conservation practices: Exotic species as an example. Environmental Management 47: 315-321.

Quezada-Euan JJG, May-Itza WJ, Gonzalez-Acereto JA. 2001. Meliponiculture in Mexico: problems and perspective for development. Bee World 82(4): 160-167.

Ramirez-Marcial N, Gonzalez-Espinosa M, Williams-Linera G. 2001. Anthropogenic distur- bance and tree diversity in Montane Rain Forests in Chiapas, Mexico. Forest Ecology and Management 154: 311-326.

Rodriguez B, Mendoza S, Iturriga MH, Castano-Tostado E. 2012. Quality parameters and antioxidant and antibacterial properties of some Mexican honeys. Journal of Food Science 71(1): 121-128.

Roseler PF. 1985. A technique for year-round rearing of Bombus terrestris (Apidae, Bombini) colonies in captivity. Apidologie 16(2): 165-169.

Roubik DW. 1980. Foraging behavior of competing Africanized honeybees and stingless bees. Ecology 61: 836-845.

Runckel C, Flenniken ML, Engel JC, Grahm Ruby J, Ganem D, Andino R, DeRisi JL. 2011. Temporal analysis of the honey bee microbiome reveals four novel viruses and seasonal prevalence of known viruses, Nosema, and Crithidia. PloS ONE 6(6): e20656.

Sandbrook C, Nelson F, Adams W, Agrawal A. 2010. Carbon, Forests and the REDD Paradox. Oryx 44(3): 330-334.

Sande SO, Crewe RM, Raina SK, Nicolson SW, Gordon I. 2009. Proximity to a forest leads to higher honey yield: another reason to conserve. Biological Conservation 142: 2703-2709.

112 Santas LA. 1983. Insects producing honeydew exploited by bees in Greece. Apidologie 14: 93-103.

SEMARNAT 2003. Acuerdo que establece las Reglas de Operacin para el otogamiento de pagos del Programa de Servicios Ambientales Hidrologicos. Secretaria de Medio Ambiente y Recurso Naturales. Diario Oficial.

Seeley TD. 1985. Honeybee ecology. Princeton University Press, Princeton.

Schaffer WM, Zeh DW, Buchman SL, Kleinhans S, Schaffer MV, Antrim J. 1983. Compe- tition for nectar between introduced honey bees and native North American bees and ants. Ecology 64: 564-577.

Schweiger O, Biesmeijer JC, Bommarco R, Hickler T, Hulme PE, Klotz S, Kuhn I, Moora M, Nielsen A, Ohlemuller R, Petanidou T, Potts SG, Pysek P, Stout JC, Sykes MT, Tscheulin T, Vila M, Walther GR, Westphal C, Winter M, Zobel M, Settele J. 2010. Multiple stressors on biotic interactions: how climate change and alien species interact to affect pollination. Biological Reviews 85(4): 777-795.

Secretaria de Agricultura, Ganaderia, Desarrollo Rural, Pesca y Alimentacion. 2009. Servi- cio de informacion estadistica agroalimentaria (SIAP) http://www.siap.sagarpa.gob. mx.

Steffan-Dewenter I, Tscharntke T. 2000. Resource overlap and possible competition between honey bees and wild bees in central Europe. Oecologia 122: 288-296.

Taylor OR, Delgado A, Brizuela F. 1991 Rapid loss of European traits from feral neotropical African honey bee populations in Mexico. American Bee Journal 131:783.

Thompson DM. 2006. Detecting the effects of introduced species: a case study of competi- tion between A. mellifera and Bombus. Oikos 114: 407-418.

Villanueva-Gutierrez R, Moguel-Ordonez YB, Echazarreta-Gonzalez CM, Arana-Lopez G. 2009. Monofloral honeys in the Yucatn Peninsula, Mexico. Grana 48(3): 214-223.

Villanueva- Gutierrez R, Roubik DW, Colli-Ucan W. 2005. Extinction of Melipona beecheii and traditional beekeeping in the Yucatan peninsula. Bee World 86(2): 35-41.

Wilcox BA. 1984. In situ conservation of genetic resources: Determinants of minimum area requirements. In National Parks, Conservation and Development, Proceedings of the World Congress on National Parks. (eds.) JA McNeely and KR Miller, Smithsonian Institution Press, pp. 18-30.

Winfree R, Aguilar R, Vazquez DP, LeBuhn G, Aizen MA. 2009. A meta-analysis of bees’ responses to anthropogenic disturbance. Ecology 90(8): 2068-2076.

113 Winston ML. 1992. The biology and management of Africanized honey bees. Annual Review of Entomology 37: 173-193.

Chapter 6

Bacandritsos N. 2002. A scientific note on the first successful establishment of the monophlebine coccid Marchalina hellenica (Coccoidea, Margarodidae) on the fir tree (Abies cephalonica). Apidologie 33(3): 353-354.

Chazdon RL. 2003. Tropical forest recovery: legacies of human impact and natural distur- bances. Perspectives in Plant Ecology Evolution and Systematics 6(1-2): 51-71.

Dungan RJ, Turnbull MH, Kelly D. 2007. The carbon costs for host trees of a phloem- feeding herbivore. Journal of Ecology 95(4): 603-613.

Gunderson LH. 2000. Ecological resiliencein theory and application. Annual Review of Ecology and Systematics 31: 425439.

Heaney LR. 2007. Is a new paradigm emerging for oceanic island biogeography? Journal of Biogeography 34(5): 753-757.

Law R, Illian J, Burslem DFRP, Gratzer G, Gunatilleke CVS, Gunatilleke IAUN. 2009. Eco- logical information from spatial patterns of plants: insights from point process theory. Journal of Ecology 97: 616-628.

Mendel Z, Assael F, Saphir N, Zehavi A, Nestel D. 1997. Seedling mortality in regenera- tion of Aleppo pine following fire and attack by the scale insect Matsucoccus josephi. International Journal of Wildland Fire 7(4): 327-333.

SEMARNAT 2003. Acuerdo que establece las Reglas de Operacin para el otogamiento de pagos del Programa de Servicios Ambientales Hidrlogicos. Secretaria de Medio Ambiente y Recurso Naturales. Diario Oficial.

Vale TR. 1982. Plants and People: Vegetation Change in North America. Association of American Geographers, Washington, DC, 1-88.

Wada N, Murakami M, Yoshida K. 2000. Effects of herbivore-bearing adult trees of the oak Quercus crispula on the survival of their seedlings. Ecological Research 15(2): 219-227.

Walker B, Hollin CS, Carpenter SR, Kinzig A. 2004. Resilience, adaptability and trans- formability in social-ecological systems. Ecology and Society 9(2): 50-59.

114 BIOGRAPHICAL SKETCH

Heather Gamper was born in 1976 and was raised in the New Jersey suburbs. Growing up she was often confronted with the emotions of seeing her favorite natural area con- verted to other human-modified habitat. This has fueled her life’s learning in biological sciences and geography. Gamper graduated from the University of Vermont in 1998 with a B.S. in Wildlife Biology. In 2002 she earned an M.S. in Biology from Florida In- ternational University. While obtaining these degrees, Gamper has spent several years conducting field research in the Dominican Republic, Puerto Rico and Mexico. Her current dissertation research has been funded by the National Science Foundation and several University fellowships. She recently was awarded the 2011 Graduate Research and Creativity Award from Florida State University. In addition to research, Gamper has taught many college courses in laboratory, field and lecture settings on topics in- cluding Ornithology, Ecology and Geographic Information Systems. A highlight of her teaching includes presenting a course on honey bees to enthusiastic, mature learners at the Osher Life Long Learning Institute (OLLI) at Florida State University.

115