Ecology and Behavior of Two Sympatric Chameleon Species on Bioko Island, Equatorial Guinea Elliott Chiu Dr

Home , Carpet chameleon, Chameleon, Namaqua chameleon, Trioceros, Trioceros ellioti, Trioceros hoehnelii

Ecology and Behavior of Two Sympatric Chameleon Species on

Bioko Island, Equatorial Guinea.

A Thesis

Submitted to the Faculty

Drexel University

Elliott Chiu

in partial fulfillment of the

requirements for the degree

Master of Science

June 2013

ACKNOWLEDGEMENTS

First and foremost, I would like to thank my parents Serena and Edward, my brother Alvin, and my sister Melissa. Without them, I would not be here; they supported me emotionally and financially and I can truly say that this project is as much theirs as it is mine.

I would like to acknowledge Dr. Tom Butynski and Dr. Torsten Wronski for their help in developing my study during the early stages of my thesis. Originally my project focused on rare occurrences between primate and duiker feeding habits, but you made me realize that projects that I envisioned would be difficult to finish even for a PhD and helped me develop a suitable Master’s thesis.

I thank all of my friends who despite my constant weariness, never gave up on me and were a welcome respite from the long hours in the lab. Long absences can strain many relationships, but I have been lucky to know all of you. Despite many rejected offers, you never abandoned me and that is the test of a true friendship.

Of course, much of the brunt of the fieldwork was shared by many dedicated volunteers. The students of the Drexel Study Abroad program on Bioko (including Evan

Neyland, Kadeem Gilbert, Ryan Quinn, Liz Long, Beth Darby, and Baltasar), Nabil

Nasseri, Andrew Fertig, Djibrilla Ousman, and Maya Lipschutz were instrumental in the overall collection of my chameleons. Reed Power provided incredibly useful habitat and density information to me when I was unable to return the field in early 2013.

I would be remiss to forget my three amazing lab volunteers, Ellen Wildner, Lexi

Khan, and Halle Choi, who spent many hours looking through dissecting scopes and ii separating out thousands of tiny invertebrates. Not only that, you remained patient during the entire bomb calorimetry fiasco.

Special mention should be given to the graduate students of the Hearn lab Drew

Cronin, Jake Owens, and Pat McLaughlin who shared many experiences with me regarding working in the tropics and never let bureaucratic nonsense impede on my progress. Additionally, I’d like to thank Drew Cronin and Steve Hromada for supplying me with the beautiful maps that have been generated for my thesis.

Funders for my project were greatly appreciated. As mentioned before, my parents were my biggest financial support, but the Bioko Biodiversity Protection Program paid for many of my expenses in country and provided me with hot food and a

“comfortable” bed/tent to sleep in. The Garden Club of America has been extremely generous in providing funds for the acquisition of many of my research supplies. Lastly,

I would like to thank ExxonMobil for their support in conservation efforts on Bioko

Island.

Finally, I would like to thank the members of my thesis committee. Dr. Gail

Hearn has been a great mentor and advisor for me, helping me with developing the sense to work in science abroad. I would be nowhere without the help of Dr. Mike O’Connor who taught me how to approach and answer scientific questions with statistics. Dr.

Krystal Tolley advised me in all things chameleon, and was an irreplaceable member of my committee. Dr. Dan Duran provided me with much needed help in collecting and identifying the insects that were used in my study. And then the most important of all is

Dr. Shaya Honarvar, who sat with me through many drafts of my thesis and personally devoted time and energy into seeing me succeed. iii

TABLE OF CONTENTS

LIST OF TABLES……………………………………………………………………....iv

LIST OF FIGURES……………………………………………………………………...v

ABSTRACT…………………………………………………………………………….vii

1. CHAPTER 1: General Introduction………………………………………………….1 2. CHAPTER 2: A life history of two of Bioko Island chameleons, the Fea’s chameleon (Trioceros feae) and the spectral pygmy chameleon (Rhampholeon spectrum)

Introduction…………………………………………………………………….…6

Methods…………………………………………………………………………..10

Results…………………………………………………………………………....13

Discussion………………………………………………………………………..17

3. CHAPTER 3: Differential feeding strategy adaptations of Trioceros feae and Rhampholeon spectrum in Bioko Island, Equatorial Guinea

Introduction……………………………………………………………………..30

Methods…………………………………………………………………………34

Results…………………………………………………………………………..37

Discussion………………………………………………………………………40

4. CHAPTER 4: Conservation implications of diet partitioning in response to differential nutritional requirements in the Fea’s chameleon (Trioceros feae) and the African leaf chameleon (Rhampholeon spectrum) on Bioko Island

Introduction……………………………………………………………….……53

Methods…………………………………………………………………….…..57

Results……………………………………………………………………….…60

Discussion……………………………………………………………………...63

5. CHAPTER 5: Notes on the natural history of Bioko Island chameleons………….74 6. CHAPTER 6: General Conclusions………………………………………………..88

LIST OF REFERENCES……………………………………………………………....90 iv

LIST OF TABLES

1. Stomach contents of gastric lavage were dissected and identified to the level of order. The number of individual prey items is listed with the number of stomach pellets in which the individual prey was collected in parentheses………………………………22

2. Intraspecific behavioral data: Total distance traveled (TDT), average speed (AS), moving speed (MS), percent time spent moving (PTM), and feeding attempts (FA) did not differ significantly with the change among sexes within species……………..….45 3. Interspecific behavioral data: Total distance traveled (TDT), average speed (AS), and moving speed (MS) were insignificantly variable between species. Percent time spent moving (PTM) and feeding attempts (FA) differed significantly between species….46

4. Binomial models determining variables affecting activity levels: The full model included five variables (sex, species, time, temperature, and the interaction between species and sex). Subsequent models removed one sequential variable until significant changes occurred with the removal of species and time………………………….…..47 5. Available invertebrates collected along each trail are categorized by the method by which they were collected. Percent (%) composition displays what percentage of the total available invertebrate selection was made up of each taxon………………..…..67 6. Stomach contents retrieved via gastric lavage are displayed for each species along with the trail on which they were collected. Numbers of prey items are reported along with the number of chameleons from which those individuals were collected, demarcated by the parentheses…………………………….……………………..…...68

7. Electivity indices (E*) are reported for T. feae along both trails. Invertebrate taxa are first blocked by prey hardness and then by evasiveness of individual prey taxa…….69

8. Electivity indices (E*) are reported for R. spectrum along both trails. Invertebrate taxa are first blocked by prey hardness and then by evasiveness of individual prey taxa…70

9. Combined electivity indices (E*) are reported for T. feae and R. spectrum. The Nature trail and the Cascades trail were grouped to produce one E* value. Invertebrate taxa are first blocked by prey hardness and then by evasiveness of individual prey taxa…71

10. Biometric measurements that were used for each captured chameleon in 2011…....78

11. Additional biometric measurements were used for each captured chameleon in 2012 (Hopkins & Tolley, 2011)…………………………………………………….………79 v

LIST OF FIGURES

1. Four species of chameleons on Bioko Island. A) Rhampholeon spectrum perched upon tree fern branches undergoing ecdysis. Its skin is shed in one solid piece instead of flaking off in small pieces. B) Trioceros cristatus photographed on the 2008 National Geographic Expedition on Bioko Island. Photo credit: Christian Ziegler ILCP Bioko RAVE. C) Trioceros oweni is a mid-sized chameleon that was originally described from the island but its presence is not disputed. D) Trioceros feae is a mid-sized chameleon endemic to sub-montane forests of Bioko Island……………………...…..23 2. Distribution maps of R. spectrum, T. cristatus, T. feae, and T. oweni in Africa…..…..24 3. General map of study sites: The Nature trail and Cascades Trail are marked in respect to the Moka Wildlife Center (MWC) located just outside of the village of Moka. …..25 4. Density of chameleons were estimated on each trail. The Nature trail is classified as a more disturbed habitat compared to the Cascades trail…………………………….….26 5. Photos of the understory of A) Cascades trail and B) the Nature. The understory has been cleared of much of the flat broad leaf vegetation in the Nature trail whereas that secondary growth is still present along the Cascades trail………………………….….27 6. Sleeping perches utilized by T. feae during the 2011 field season. The majority of sleeping perches utilized were exotic species. …………………………………...……28 7. A sub adult male T. feae and adult male R. spectrum are shown here shortly after being aroused by observers in the forest. A) The T. feae is utilizing the small graspable twigs and branches to perch upon whereas B) R. spectrum use flat broad leaves to rest upon. They typically are not found to grasp small perches…………………………………..29 8. Differences in total distance traveled is plotted with respect to sex within each species as a result of direct 12-hour observation cycles (n=5). There are no statistical differences between species or sex……………………………………………….……48

9. Differences in average speed is plotted with respect to sex within each species as a result of direct 12-hour observation cycles (n=5). Differences between sexes in R. spectrum are suggestive of statistically significance…………………………………..49

10. Differences in moving speed is plotted with respect to sex within each species as a result of direct 12-hour observation cycles (n=5)………………………...……………50

11. Differences in percent of time spent moving is plotted with respect to sex within each species as a result of direct 12-hour observation cycles (n=5). Differences between species are statistically significant……………………………………………………..51

12. Differences in feeding attempts is plotted with respect to sex within each species as a result of direct 12-hour observation cycles (n=5). Differences between species are statistically significant…………………………………………………………..……..52 vi

13. Electivity indices (E*) were graphed to compare A) prey evasiveness and B) prey hardness between R. spectrum and T. feae, combining data from the Nature and Cascades trails……………………………………………………………………….....72 14. Population structure is given as a percentage of total T. feae chameleons captured during the 2011 (n=79) and the 2012 (n=100) field seasons…………………………..73

15. Population structure is given as a percentage of total T. feae chameleons captured during the 2011 (n=79) and the 2012 (n=100) field seasons………………………….80

vii

ABSTRACT Ecology and behavior of two sympatric chameleon species on Bioko Island, Equatorial Guinea Elliott Chiu Dr. Gail Hearn, Advisor, PhD.

Chameleons are curious creatures that may be found in many diverse habitats all throughout the Old World. The overwhelming diversity found in places such as

Madagascar is a testament to the different behavioral and morphological adaptations that chameleons have incurred that allow them to survive with limited competition. The survival of Rhampholeon spectrum and Trioceros feae in a given habitat on Bioko Island allow for the comparison of different factors that may affect the ecology of each chameleon. Analysis of sleeping behavior and habitat differences along two trails on

Bioko Island suggest that R. spectrum may not survive as easily as T. feae in disturbed habitats due to the removal of flat broad-leafed vegetation in the understory.

Feeding ecology of chameleons has been generally understudied for some time until recently. While other lizard families can easily be classified into the sit-and-wait and active forager paradigm, chameleons do not easily fit the mold. Direct observations of daily behavior showed variations in the percent of time spent moving and feeding attempts between species, while also suggesting gender-based differences in behavior in

R. spectrum chameleons.

Stomach contents retrieved using gastric lavage of 103 chameleons revealed differential resource allocation of available prey. Trioceros feae preyed upon Coleoptera and Diptera more heavily that R. spectrum where as Araneae and Orthoptera were preyed viii upon more heavily by R. spectrum. Electivity indices suggest that prey selection was not based on prey hardness or prey evasiveness but rather other factors entirely. 1

CHAPTER 1: General Introduction

The family Chamaeleonidae is represented by more than 195 species within 11 genera (Tolley et al., 2013). Laterally flattened bodies, ballistic tongues, prehensile tails, forceps-like feet, independent stereoscopic eyes, and most notably, the ability to change colors distinguish chameleon morphology (Necas, 1999). Due to these unique characteristics, chameleons have always fascinated scientists and laypeople alike, generating a great desire to propagate further research. While many species of chameleons have consistently been the center of the scientific spotlight in terms of research, new species of chameleon are constantly being discovered (Necas et al., 2003;

Gehring et al., 2010; Glaw et al., 2012).

The earliest known fossils dating from the Miocene epoch (23-5 mya) came from the Czech Republic and more recently, Southern Germany, but they are not believed to represent the earliest chameleons (Moody & Rocek, 1980; Čerňanský, 2011). The origin of chameleons has been debated. Initially, chameleons were believed to have originated from East Africa, after the split of Africa and Madagascar (Hillenius, 1959).

Researchers studying morphology situated the origin of chameleons in ancient Gondwana

(Klaver & Böhme, 1986). A combination of immunological distances and calibrated albumin molecular clocks supported a Gondwanan origin, though others believe that the claim was erroneous, unsupported by modern molecular data (Hoffman et al., 1991;

Rieppel, 2002). Further morphological studies reestablished a post-Africa-Madagascar split with the origin of the family Chamaeleonidae in Madagascar (Rieppel & Crumly,

1997). 2

More sophisticated mtDNA, NADH, and tRNA analyses supported a post-

Gondwanan origin and a reorganization of the family Chamaeleonidae (Townsend &

Larson, 2001; Raxworthy, 2002). Paleologically speaking, the oldest chameleon fossils have been dated back to the Miocene epoch (26 mya); although lizard fossils, let alone chameleon fossils, are poorly represented in the fossil record, typically leaving isolated cranial bones (Moody & Rocek, 1980; Rieppel et al., 1992; Raxworthy et al., 2002).

Recent research supports a theory that chameleons diverged from their earliest common ancestors in the mid- to late-Cretaceous (65-100 mya), much earlier than the dated origin of the earliest fossils found (Townsend et al., 2011b; Tolley et al., 2013). The most widely cited work to date is that chameleons first evolved in Madagascar, where diversity is greatest housing more than 60 species on the island, with new species being described as recent as 2012 (Black et al., 2000; Yoder & Nowak, 2006; Gehring et al., 2010; Florio et al., 2012; Crottini et al., 2012; ). According to this theory, chameleons radiated out of

Madagascar on multiple occasions to continental Africa, and the Seychelles and Comoros

Islands, via oceanic dispersal (Raxworthy et al., 2002; Rieppel, 2002).

This theory loses support with its lack of molecular data. Recently, studies have hypothesized an African dispersal of chameleons that makes up where the Malagasy theory falls short (Townsend et al., 2011a; Tolley et al., 2013). The study analyzed three ancestral-state reconstruction models that would support chameleon phylogeny by maximum parsimony using modern genetic techniques. Using 6-gene and 13-gene datasets, the researchers were able to adequately support genus classifications, though the analyses differed in the resolution of deeper nodes of the tree. A 6-gene BEAST relaxed- clock analysis provided better resolution supporting some classification beyond the level 3 of genera. Mitochondrial and nuclear markers show rapid diversification of the chameleon family, however, the chameleons appear to have diverged from other reptiles during the Late Cretaceous (65 – 100 mya) with genera differentiating during the Eocene

(34 – 56 mya). While this may be the case, this theory does not dispose of the argument that Malagasy radiations occurred at all. In fact, the chameleons inhabiting the Comoros archipelago are more closely related to those found on Madagascar (Townsend et al.,

2011a). The Indian/ Sri Lankan species however, originated from an African stock

(Tolley et al., 2013). The theory is further supported by the fact that the sister taxa of

Chamaeleonidae, Agamidae, does not occur on Madagascar, nor the Comoros and

Seychelles archipelago (Tolley et al., 2013). However, both of these families of lizards can be found living together in much of the rest of Africa as well as other islands such as

Bioko Island (Eisentraut, 1973).

It is believed that ocean dispersal was important for the dispersal of few chameleon groups (Raxworthy et al., 2002). While dispersion on rafts of driftwood or other buoyant materials is rare, it appears to have been key for 2 groups of Malagasy chameleons from Africa, and subsequently, for the Comoros Island species from

Madagascar (Tolley et al., 2013). Bioko Island is a small volcanic island with an area of

2,017 km2 found 32 km off the coast of Cameroon that separated from the mainland approximately 12,000 years ago, appears to have been populated by other means. With such a recent isolation, it is likely that the chameleon species that can be found on the island were originally located on the island before the land bridge disappeared (Lee et al.,

1994). Four species have been found on Bioko Island: Rhampholeon spectrum, Trioceros cristatus, T. oweni, and T. feae (Larison et al., 1999; Necas, 1999; Barej et al., 2010; 4

Tilbury, 2010). Collectively, these four species inhabit sub-Saharan West Africa ranging from Nigeria East to the Democratic Republic of Congo and South into Angola (Mariaux

& LeBreton, 2010; Tilbury, 2010; Carpenter, 2011; LeBreton et al., 2011).

Of the four chameleon species found on Bioko Island, T. feae and R. spectrum remain elusive and have yet to be described in detail. While it exists in other localities outside of Bioko Island, R. spectrum has not been the focus of many studies. The lack of threat to R. spectrum populations, their wide distribution, small stature, and elusive nature do not make them particularly charismatic in respect to other species of chameleons

(Mariaux & LeBrenton, 2010). R. spectrum is the only species of the genus found on the

Western side of the continent, having separated from the other lineages during the mid- to late- Oligocene Epoch (26.19 ± 7.49 mya) (Mathee et al., 2004). Trioceros feae, on the other hand, is an endemic species found only in the submontane-montane forests of

Bioko Island, known to inhabit elevations between 1,300 and 1,600 m above sea level

(Carpenter, 2011). This medium-sized chameleon is poorly represented in scientific literature (Klaver & Böhme, 1992; Tilbury & Tolley, 2009; Barej et al., 2010; Tilbury,

2010). While population trends are unknown, threats to the species include habitat destruction and harvesting for the pet trade. General ecological information is missing and there is an urgent need for research regarding many aspects of its life history

(Carpenter, 2011). Phylogenetically speaking, T. feae is most closely related to T. camerunensis, T. montium, and T. cristatus, all of which can be found in neighboring

Cameroon (Pook & Wild, 1995).

The general aim of my thesis is to fill the gaps in scientific literature that remain for both R. spectrum and T. feae. In Chapter 2, I attempt to provide life history, ecology, 5 and behavioral traits for both species with the purpose of comparing the two species and highlighting information that is previously unknown for T. feae. In chapter 3, I explore the differences in feeding strategies between R. spectrum and T. feae in respect to foraging behaviors. In chapter 4, I examine differential feeding strategies and resource allocation of available prey as a response to human-related habitat disturbance. In chapter 5, additional data collected that were not used to address particular questions or hypotheses are provided. Finally in chapter 6, I will reassess my findings in order to provide previously unknown information about the life history, ecology, and behavior of a poorly-studied chameleon and address the implications for its survival in disturbed habitats.

CHAPTER 2: A life history of two of Bioko Island chameleons, the Fea’s chameleon (Trioceros feae) and the spectral pygmy chameleon (Rhampholeon spectrum)

INTRODUCTION

Due to their extensive distribution throughout Africa, Madagascar, Southern

Europe, the Arabian Peninsula, India, and Sri Lanka, chameleons occupy a wide variety of habitats. In Africa alone, there is great habitat diversity: some chameleons are montane species that can survive in cool environments such as T. pfefferi, T. quadricornis, and T. wiedersheimi in various Cameroonian mountain ranges; while others are lowland rainforest species that experience constant rains like T. cristatus on Bioko Island and T. oweni in Western Africa; others still are restricted to arid deserts such as Chamaeleo namaquensis in the Southwestern African deserts of Angola, Namibia, and South Africa

(Necas 1999; Barej et al., 2010; Tilbury, 2010). The habitat differences experienced throughout their range have imposed various physiological constraints that drive adaptive radiation (Tilbury, 2010).

Among the adaptations found in chameleons are differences in morphology and size. Many chameleons that have adapted to an arboreal life have longer prehensile tails and longer limbs; while chameleons that are terrestrial tend to have shorter tails, shorter limbs, and poorer grasping performance due to smaller hands (Bickel & Losos, 2002;

Herrel et al., 2013). There exists a vast diversity of chameleon sizes; the smallest chameleon, Brookesia micra, is 15 mm in length while the largest chameleon, Furcifer oustaleti, can reach lengths of up to 68.5 cm (Glaw & Vences, 2007; Necas, 1999). 7

The population structure and density of chameleons are also quite variable. While originally believed to be universally solitary, observations have shown that the density of chameleons is varied among species, with some living in dense populations (e.g. Furcifer pardalis, F. lateralis, and T. affinis) (Necas, 1999). There has also been one instance in which researchers documented on long-term monogamous mating pair in T. hoehnelii

(Toxopeus et al., 1988). Reproductive biology is also quite variable among different species of chameleon. Gestation periods range from 20 days in Chamaeleo chamaeleon to 236 days in Trioceros jacksonii and Furcifer campani. The time between oviposition and mating is variable but may be as short as 14 days in Trioceros ellioti. The number of eggs laid may be as few as 2 eggs per reproductive season in Rhampholeon spectrum and may even reach up to a maximum of 93 eggs in Chamaeleo chamaeleon (Necas, 1999).

Chameleons can be considered, in part, social animals and have developed ways of communicating with each other. Physical aposematic behaviors such as “pumping” their bodies or inflating the buccal cavity may often be used in lieu of physical combat

(Tokarz et al., 2005). Physical mating displays will often be accompanied by color displays. Despite popular belief, chameleons do not typically used color change as a form of camouflage. The intrinsic green or brown colors of chameleons often serve as suitable camouflage. Instead, the ability to change colors is useful in thermoregulation and expression of stress as well as communication (Walton & Bennett, 1992; Necas, 1999;

Bennet, 2004; Stuart-Fox & Moussalli, 2008). Popular examples thermoregulation and warning coloration display can be seen in the Namaqua chameleon (Chamaeleo namaquensis) and in the veiled chameleon (C. calyptratus), respectively (Tolley &

Burger, 2007; Necas, 1999). 8

The spectral pygmy chameleon, R. spectrum is one of 14 extant species that belongs to one of the oldest clades of chameleons known as African leaf chameleons

(Genus = Rhampholeon; Fig. 1A). It is also considered a false chameleon, differing from other groups in the way they shed their skin. While most chameleons undergo ecdysis producing patchy flakes, R. spectrum is among the few chameleons that produce a relatively intact exuvium (Necas, 1999). The genus consists of predominantly East

African species with the exception of R. spectrum, which can be found in Congo and

Cameroon as well as on Bioko Island from elevations spanning from sea level to 1,900 meters (Fig. 2) (Necas, 1999; Mathee et al., 2004; Mariaux & LeBreton, 2010; Tilbury,

2010). R. spectrum is a small chameleon typically not exceeding 10 cm with a short prehensile tail. The species is brown in color and can be found in forested areas in close proximity to the ground and among low shrubbery. Males and females are sexually dimorphic and males are distinguished by the enlarged base in which the hemipenes is ensheathed (Necas, 1999). Little is known of their behavior and feeding habits (Tilbury,

2010).

The other three chameleon species found on Bioko Island belong to the genus

Trioceros (originally classified as a subgenus under the genus Chamaeleo) ( laver &

B hme 1986). Previous classification schemes were based on reproductive and lung morphology ( laver & B hme 1986). Re-evaluation of the genus Chamaeleo was prompted by new molecular phylogenetic and mitochondrial genomic evidence

(Townsend & Larson, 2002). The new classification is organized based on both morphological (hemipenal, lung, and skull morphology) and genetic data. Analysis of 9 two mitochondrial genes and one nuclear gene supported the move of 36 Chamaeleo species to the newly established genus, Trioceros (Tilbury & Tolley, 2009).

Trioceros cristatus can be found in Equatorial Guinea, Cameroon, the Central

African Republic, the Democratic Republic of the Congo, Congo, Gabon, Nigeria, Ghana and Togo (Fig. 2). It is a medium-sized chameleon (15-30cm) with a maximum total length of 28 centimeters. It is one of the most easily recognizable species of chameleon with a unique row of blue scales found along its casque (Fig. 1B) (LeBreton et al., 2011).

Trioceros oweni is also a West African species that was originally collected and described from the type specimen collected in 1831 on Bioko Island (Fig. 1C) (Grey,

1831). However, since the original description, it has not been seen on the island by scientists, collectors, or local people. It is possible that the original specimen was erroneously attributed to Bioko Island, a situation that has arisen with other biological specimens being transported back from West Africa on British ships. Bioko Island (then

Fernando Poo) was a major British port (then Port Clarence and now Malabo) throughout most of the 1800’s.

Bioko Island is only home to one endemic species of reptile, T. feae (Fig. 1D)

(Klaver & Böhme, 1992). It was previously treated as a subspecies of T. montium, but was later postulated to be T. montium’s closest extant relative (Necas, 1999). Trioceros feae was described from one location on the island distributed at an altitude of 1300-1600 m around the village of Moka (Carpenter, 2011). Aside from its distinguishing external, hemipenes, and lung morphology, little else is known of this medium-sized submontane species (Klaver & Böhme, 1992). 10

Both T. feae and R. spectrum were included in this study because of their overlapping ranges at higher elevations. The most striking difference between T. feae and R. spectrum lies in the classification between the two species. Rhampholeon spectrum is much smaller, brown in color, and lives in the understory of forests.

Members belonging to Chamaeleoninae are typically larger, are green in color, and are predominately arboreal species (Tilbury, 2010). The purpose of this study is to determine what biological characteristics allow sympatric species to coexist. Given their differences, it is likely that while the two species inhabit the same habitats, they occupy different niches and microhabitats and will also have different behaviors and different diets.

METHODS

Study site

Data collection took place between the months of February -May 2011 and the

March – June 2012. Two field sites spanning elevations from 1,100m to 1,500m included the Nature Trail (N 03˚21’41.3”, E008˚39’44.5”) and the Cascades Trail (N

03˚19’31”, E 008˚40’297”) in Moka, Bioko Island, Equatorial Guinea (Fig. 3). Data was collected during the February – May 2011 and the March – June 2012 field seasons.

Density

Chameleon density data was obtained by sampling quadrants along both the

Nature Trail and the Cascades Trail during nighttime surveys. In an attempt to determine the chameleon density of T. feae and R. spectrum, five 100-meters transect lines were created running parallel to one another other, 5 meters apart on both the Nature and 11

Cascades Trails. For consistency, survey points were flagged with tape every 5 meters along the 100– meters transect. At each mark, chameleon searches were performed on both sides of the transect line for 2.5 meters to avoid resampling.

Species, time, meter mark, altitude, and the GPS coordinates were recorded at each chameleon sighting. The total density of chameleons (individuals/625m2) was estimated using the following equation:

whereby “ ̅” represents the total number of chameleons sighted per trail, and “A” represents the total area of surveyed for each transect (625 m2).

Habitat

Habitat assessments were conducted using the same transects as described above along the same trails during the daytime. At each meter mark, canopy closure (%) was recorded using a densitometer, visibility (m) was measured by determining the distance an observer can see through the understory at eye level, substrate consistency was recorded marking the type of substrate (e.g. dirt, leaf litter, rocks), substrate depth (mm), and plant species were also documented.

Sleeping habitat selection notes were recorded during the 2011 field season from

T. feae only. During each chameleon collection marking perch height, perch diameter, and perch species whenever possible.

Behavior 12

During the late dry and early wet seasons (April 17 – June 8, 2012), behavioral observations were run for 7 Rhampholeon spectrum males, 5 R. spectrum females, 5

Trioceros feae males, 5 T. feae females. Inter- and intraspecific interactions and color changes were recorded every 5 minutes for twelve hours along the Nature Trail.

Observers maintained a distance of at least 3 m, hid behind cover wherever available, and used binoculars whenever necessary so as not to disturb normal chameleon behavior.

Color changes were regarded as a proxy for observer interference. No chameleon was resampled during the study. While left unmarked, the chameleons were identified by their location and unique natural markings.

Diet

A total of 200 chameleons consisting of 100 R. spectrum and 100 T. feae chameleons (50 of each species per trail) were collected and stomach contents were obtained via gastric lavage. A rigid plastic ring was inserted into the chameleons’ mouths to ensure that the chameleon would not close its jaws during the procedure. A plastic catheter was fed into to stomach of the chameleon. 100cc of water was injected into the stomach until contents of the chameleons were flushed out and captured by sieve

(Measey et al, 2011). Once obtained, stomach contents were weighed to determine their wet weight before being preserved in 95% ethanol and until they were identified to the level of order using Johnson & Triplehorn (2004). Dietary analysis was performed examining differences in respect to species and sex of the chameleons.

RESULTS

Density

A total of 212 chameleons (76 T. feae and 145 R. spectrum) were sighted on the

Cascades trail (sampling days, n=3) and the nature trail (sampling days, n=7). The average number of chameleon sightings per night of T. feae on Nature trail and Cascade trail was 9.285 and 0.667, respectively. The average number of chameleon sightings per night of R. spectrum on the Nature trail and Cascade trail was greater than the average number of T. feae on both accounts with values of 12 and 20.333 respectively. Using

Equation 1, the density of chameleons was estimated. The density of T. feae along the

Nature trail was 0.0149 chameleons/m2 (approximately 9 chameleons per 625m2) and

0.00107 chameleons/m2 (approximately 1 chameleon per 6252) along the Cascades trail.

Densities of R. spectrum were greater along both trails. R. spectrum chameleons could be found at a density of 0.0192 chameleons/m2 (12 chameleons/625m2) and 0.0325 chameleons/m2 (approximately 20 chameleons/625m2) along the Nature trail and

Cascades trail, respectively (Fig. 4).

Habitat

A total of 500 meters2 were surveyed in all transects on both the Nature and

Cascade trails. The Nature trail scored an average canopy cover of 81.286  0.732%

(mean  SE) with a minimum canopy coverage of 60% and a maximum canopy coverage of 95%. Visibility averaged 12.048  0.544 meters (mean  SE) with a minimum and maximum visibility of 5 and 20 meters, respectively. Leaf litter and dirt composed the majority of substrate, followed by saplings and then by rocks. Substrate depth ranged from 5 to 40 millimeters with an average of 19.24  0.92 mm. 14

The Cascade trail averaged 82.714  0.628% canopy coverage with a minimum canopy coverage of 60% and a maximum canopy coverage 90%. Visibility averaged

7.857  3.526 meters with a minimum and maximum visibility of 5 and 15 meters, respectively. Substrate was predominately composed of dirt and shrubs. Primary vegetation consisted of Aframonum spp. and Cyathea Camerooniana. Canopy coverage on the Nature trail averaged 81.3 ± 0.7% whereas canopy coverage along the Cascades trail averaged 82.7 ± 0.6%. Differences in canopy coverage were not significant between the two trails (t=1.482; df=203.263; p=0.14) suggesting primary growth was relatively similar between the two trails. Visibility along the Nature trail averaged 12.0 ±

0.7m and 7.9 ± 0.3m along the Cascades trail. Differences in visibility were significantly different between the trails (t=6.508; df=175.682; p<0.001). These results suggest clearing of low- to mid- level vegetation within the forest. This is often done to allow for growth of shade-tolerant crops (Fig. 5A– B).

Sleeping habitats were recorded at the time each chameleon was collected in

2011. Chameleons were found to perch on various species of plants; 35 individuals were found sleeping on Pennisetum purpureum, 8 on Prunus africana, 6 on Pteridium aquilinum, 4 on Trema orientalis, 2 on Croton ssp., 2 on vines, 5, on exotic species, and

15 on unidentified species (Fig. 6). Due to a lack of vegetation identification experience,

R spectrum sleeping perches were left unidentified. Sleeping height above ground ranged from 0.5m to 5m above the ground across all specimens. Perch diameter recorded from both seasons ranged from 0.1 mm to 11.5 mm with an average and SE of 2.5 ± 0.1 and graspable perches were utilized by T. feae. While T. feae physically grasped onto lianas, 15 petioles, and small branches, R. spectrum could be found on top of the leaf of the plant itself (Fig. 7A–B).

Behavior

Ten observations were completed for T. feae. Among all the observations, deviations from the default color green were noted. In all but one case, color changes were common in the morning between the hours of 6:30 – 10:00. In addition, color changes were often characterized by brown coloration, aside from one case. The exception was a female that turned in black color early in the morning and then maintained a brown color for the duration for the day.

There was one noted case of an interspecific interaction; however this was a predatory event. As an adult male T. feae was travelling on the ground to another shrub, a shrew of unknown species unsuccessfully attacked the male. The male reared back and bit the shrew in defense, causing it to abandon its goal.

Intraspecific interactions consisted of “head bobs” between males and females of

T. feae for 5 separate individuals. These interactions resulted in one unsuccessful copulation attempt with a female that was not a subject of the observation.

Observational data for R. spectrum was less informative. As their default color is brown, it was very difficult to distinguish changes in color. The individuals also did not tend to interact with other individuals of the same species during times of observation, nor were there recorded predation events.

Diet

Stomach contents could not be retrieved from 97 of the 200 chameleons that underwent gastric lavage. Fewer chameleons (16 R. spectrum and 22 T. feae 16 chameleons) on the Nature trail yielded stomach contents, possibly an artifact of the mild learning curve associated with the technique. Cascades trail gastric lavages were more successful as 31 R. spectrum and 35 T. feae chameleons yielded results.

Both species fed predominately upon insects; however, R. spectrum diets consisted of a greater number of spiders (order = Araneae). Insect orders most heavily preyed upon included Coleoptera, Diptera, Hemiptera, Hymenoptera, Lepidoptera larvae, and Orthoptera (Table 1). There were two non-arthropod exceptions located within the chameleon gut contents. One gastropod was found within the stomach contents of a male

T. feae. R. spectrum likewise fed predominately upon arthropod species with the exception of one individual that fed upon seeds from the family of Apocynaceae. R. spectrum chameleons captured from the Nature trail collectively ate a total of 24 food items; whereas those captured on the Cascade trail ate a total of 69 items. Likewise, T. feae chameleons fed upon more prey along the Cascade trail than those captured from the

Nature trail; those captured along the Nature trail collectively fed upon 61 prey items, whereas those captured on the Cascade trail ate a total of 110 arthropods. Per chameleon,

R. spectrum’s stomach contents contained an average of 1.979 ± 0.173 prey items per stomach pellet. This differed significantly from the 3.000 ± 0.206 prey orders found in T. feae’s stomach (t=3.7887; df= 101.429; p<0.001).

DISCUSSION

Density

It was once believed that all chameleons respond negatively to habitat disturbance, making them possible bio indicators of terrestrial forest degradation; however, recent work has shown that chameleons vary in their response to habitat alteration (Necas, 1999; Jenkins et al., 2003; Patrick et al., 2011; Shirk & Vonesh, 2013).

This is further evidenced by the two species included in this study. The density of the two species appears to differ between the two sites sampled. R. spectrum chameleons appeared to thrive better in habitats that were more pristine and had a denser understory.

As small chameleons that tend to stay in closer proximity to the forest floor than T. feae,

R. spectrum seemed to benefit from the protection.

This is in stark contrast to T. feae, which could be found at a whole order of magnitude denser along the Nature trail (deemed as the more fragmented and disturbed habitat of the two). There are a couple possibilities that may give insight into this observation. T. feae may thrive better in less dense understory habitats as they are larger individuals that tend to move laterally throughout their environment. It is also possible that T. feae is a forest edge species, benefiting from ecotones. If this were true, habitat fragmentation may positively affect chameleon populations, providing more habitats as opposed to destroying suitable habitat for T. feae to inhabit.

The majority of R. spectrum chameleons was observed to have slept on top of vegetation and leaves themselves as opposed to grasping onto perches with small diameters that would be appropriate for more arboreal species. The removal of understory vegetation that provides a wide base for R. spectrum on which to sleep may 18 inadvertently significantly reduce the area of suitable habitat, thereby reducing the survivability of R. spectrum in disturbed habitats.

Other explanations may include sampling bias. Rhampholeon spectrum may be easier to locate the T. feae. In order to remove effects of human error, detections probabilities must be established before a true density can be assessed. Chamaeleoninae also utilize their volumetric habitats to a greater extent when compared to members of the genus Rhampholeon, which tend to stay closer to the forest floor. One final possibility is that fragmentation of the environment may bolster their prey populations, providing an environment with more densely available prey. In order to more accurately assess habitat degradation effects on chameleon density, intraspecific studies must be completed across a variety of degraded habitats.

One concern with the methods used in assessing chameleon density is the possibility of missing some individuals upon observation, particularly T. feae chameleons. While R. spectrum are terrestrial chameleons, rarely ascending more than a meter and a half off of the ground, T. feae can ascend into the forest canopy.

Additionally, some forests are very dense with foliage density that would obstruct visibility. However, maintaining an active searching distance of 2.5m greatly reduces the possibility of missing chameleons along a particular transect.

Habitat

The Cascades trail is located 4 kilometers SSE from the Nature trail, which is in closer proximity to the village of Moka and suffers from forest clearance and rampant insecticide usage. Village agricultural practices involve wide and indiscriminant forest clearance, which is likely the reason for significantly different visibility between both trails. These practices do not include expansive clearance of forest and so the canopy 19 coverage would not be vastly different. Furthermore, the Nature trail’s substrate is predominantly leaf litter and saplings. The understory vegetation is often cleared to make way for shade-tolerant crops, and so there is less dense vegetation available in this habitat.

The Cascades trail on the other hand, is considered to be a less disturbed habitat.

A denser forest with lesser visibility could potentially be a more suitable habitat for chameleons, allowing them to more successfully elude predation.

Native vegetation did not appear to affect the sleeping habits of T. feae as the majority of the collected chameleons could be found on invasive African elephant grass

(Pennisetum purpureum). Other factors including density (see below) suggest that T. feae’s survivability is not negatively affected by human disturbance.

Behavior

Previous studies have shown that chameleons may utilize their color changing abilities to increase, or decrease, solar radiation absorption (Necas, 1999). While this theory seems plausible and even heavily utilized in species that live in forest edges or in scrub brush habitats, it does not seem to be as useful for forest-understory species such as

R. spectrum. Other Trioceros and Chamaeleo species will climb into the canopy to bask.

While the T. feae chameleons surveyed did show evidence of early morning color changes, it would appear that the color changes were reserved mostly for communication with other nearby chameleons.

Another form of communication witnessed was the dewlap display, characterized by “head bobs” and thoracic pumping. Dewlap displays were completed when chameleons were in close proximity of other chameleons. These communication events 20 can either be interpreted as recognition of another individual, courtship, or even defensive territorial displays (Tokarz et al., 2005).

Diet

Prey targeted by T. feae and R. spectrum seems to represent different orders of invertebrates. Arthropod orders such as Diptera and Coleoptera appear to be fed upon more heavily by T. feae, whereas other orders such as Araneae and Orthoptera can be found in more R. spectrum stomach pellets. These orders may represent a difference in prey availability within a given niche, or they may indicate a larger ecological consequence of behavior. As the diversity of orders that are fed upon seem to be consistent between both species of chameleons, it does not seem likely that availability of prey is significantly different.

On the other hand, there is potential evidence that larger ecological consequences may be at play. Differences in behavior, specifically in terms of activity, may induce differential energy balance pressures upon the chameleons, necessitating high caloric intakes in T. feae chameleons. This can be achieved in two ways; either the chameleons eat more prey, or they can feed upon higher-quality prey. The quality of prey is open to interpretation but may be regarded as the food that provides the most energy.

Nutritionally speaking, food needs to satisfy both macronutrient and micronutrient requirements. While the requirements of carbohydrates, protein and fat intake may be important, the total caloric intake may prove to be more important for an ectotherm.

Diptera and Coleoptera represent orders that are higher in fat content in comparison to

Araneae and Orthoptera, and are thus more energy dense (Norberg, 1978; Barker et al.,

1998; Finke, 2002). The importance of micronutrients has also been discussed in respect 21 to diet choice, but is very difficult to study. While macronutrients may be easier to examine, caloric content is by far the crudest nutritional statistic and may prove useful in understanding how chameleons feed and what they feed upon.

It is also important to note that invertebrates retrieved from the stomachs of chameleons represent a very small sample of the available diversity in the habitat.

Sampling occurred during a relatively brief period of time between the dry and wet seasons on Bioko Island and different species or even different orders may be present at different times of the year. It may be possible that during different seasons, T. feae and

R. spectrum feed upon different groups of invertebrates.

Preliminary studies of T. feae and R. spectrum ecology and behavior show differential habitat usage and non-overlapping realized niches in regards to sleeping behavior and diet, suggesting low levels of competition between the two species. Further research needs to be done to adequately determine and establish chameleon density and continued surveys will help to address changing population dynamics from year to year.

Only then can needs for protection be assessed.

Table 1. Stomach contents of gastric lavage were dissected and identified to the level of order. The number of individual prey items is listed with the number of stomach pellets in which the individual prey was collected in parentheses.

Nature Trail Cascades Trail Total R. R. R. spectrum T. feae spectrum T. feae spectrum T. feae Order (n=16) (n=22) (n=31) (n=34) (n=47) (n=57)

Gastropoda 1 (1) 0 1

Diplopoda 4 (4) 4 (3) 4 4 Isopoda 1 (1) 1 0

Araneae 3 (3) 4 (4) 18 (16) 7 (6) 21 11

Collembola 1 (1) 0 1

Coleoptera 1 (1) 13 (11) 4 (4) 24 (19) 5 37 Diptera 8 (5) 22 (15) 13 (11) 22 (21) 21 44 Hemiptera 2 (2) 10 (9) 11 (9) 21 (17) 13 31 Hymenoptera 1 (1) 8 (6) 6 (5) 15 (11) 7 23 Lepidoptera 1 (1) 1 (1) 10 (8) 1 11 Neuroptera 1 (1) 1 (1) 0 1 Orthoptera 7 (7) 1 (1) 9 (8) 2 (2) 16 3 Unknown 2 (2) 2 (2) 3 (3) 4 3

Seed 1 (1) 1 0 Total 24 61 69 110 93 171

A B

C D A A

Figure 1. Four species of chameleons on Bioko Island. A) Rhampholeon spectrum perched upon tree fern branches undergoing ecdysis. Its skin is shed in one solid piece instead of flaking off in small pieces. B) Trioceros cristatus photographed on the 2008 National Geographic Expedition on Bioko Island. Photo credit: Christian Ziegler ILCP Bioko RAVE. C) Trioceros oweni is a mid-sized chameleon that was originally described from the island but its presence is not disputed. D) Trioceros feae is a mid-sized chameleon endemic to sub-montane forests of Bioko Island.

Figure 2. Distribution maps of R. spectrum, T. cristatus, T. feae, and T. oweni in Africa. 25

Figure 3. General map of study sites: The Nature trail and Cascades Trail are marked in respect to the Moka Wildlife Center (MWC) located just outside of the village of Moka.

5 Density Density (chameleons/625m2) 0 T. feae T. feae R. spectrum R. spectrum (Nature Trail) (Cascades Trail) (Nature Trail) (Cascades Trail) Species (Trail)

Figure 4. Density of chameleons were estimated on each trail. The Nature trail is classified as a more disturbed habitat compared to the Cascades trail.

Figure 5. Photos of the understory of A) Cascades trail and B) the Nature. The understory has been cleared of much of the flat broad leaf vegetation in the Nature trail whereas that secondary growth is A still present along the Cascades trail.

Pennisetum purpureum Vines Trema orientalis Prunus africana Pteridium aqualinum Croton ssp. Exotic ssp. Unidentified ssp.

Figure 6. Sleeping perches utilized by T. feae during the 2011 field season. The majority of sleeping perches utilized were exotic species.

Figure 7. A sub adult male T. feae and adult male R. spectrum are shown here shortly after being aroused by observers in the forest. A) The T. feae is utilizing the small graspable twigs and branches to perch upon whereas B) R. spectrum use flat broad leaves to rest upon. They typically are not found to grasp small perches.

CHAPTER 3: Differential feeding strategy adaptations of Trioceros feae and Rhampholeon spectrum Bioko Island, Equatorial Guinea

INTRODUCTION

Researching an organism’s feeding ecology allows investigators to better understand possible mechanisms that drive the survival of a species. If an organism’s attempt to maximize net energy increases fitness, feeding strategies may contribute to fitness (optimal foraging theory) (MacArthur & Pianka, 1966; Schoener, 1971). One strategy to optimize foraging success is to pursue resources that are more closely situated in proximity, thereby reducing energy expended as a result of foraging (MacArthur &

Pianka, 1966). Another strategy is to forage for a food item that does not require long handling times. Longer handling times may have a significant effect on the total amount of food that can be processed in a given foraging period (Saarikko, 1989). For instance, fig wasp eggs experienced lower mortality when oviposited within the inner layers of fig ovules, where seeds required less handling time, then when oviposited at the periphery

(Wang et al., 2013). Another common foraging strategy is cooperative foraging, which can improve fitness for social carnivores, and to a lesser extent, social omnivores

(MacDonald, 1983; Tennie et al., 2009; Bailey et al., 2013). Though sharing a meal might reduce the amount of available energy per individual, hunting in packs allows organisms to tackle larger, more dangerous prey while simultaneously reducing the risk of injury and energetic costs and ultimately, increasing energy intake per individual

(Bailey et al., 2013).

Classically, four key elements are considered when studying feeding strategies: 1) diet, 2) foraging space, 3) foraging period, and 4) foraging-group size (Schoener, 1971). 31

Traditionally, feeding strategies were separated into two main categories, the “sit-and- wait” and the “active” foraging approaches (Schoener, 1971; McLaughlin, 1989). No single strategy is optimal for all species; organisms evolve different strategies that work best with their life history (Anderson & Karasov, 1981). Energy gained using the sit-and- wait paradigm is simply calculated by the energy gained from the prey item divided by the handling time. In the active foragers, energy gained is decreased by the energy expended in searching for prey items; prey encounter rates for active foragers are much higher than for sit-and-wait predators (Schoener, 1971).

Most research on lizard feeding strategies has been done predominately on two lizard families, Iguanidae and Lacertidae and remains to be examined for many other families (Butler, 2005). Lizards either wait until prey moves within their field of attack

(sit-and-wait predator), or they spend time and energy moving from location to location, hunting sedentary prey or other easily captured prey (active foragers) (Huey & Pianka,

1981). Prey capture modes correspond to morphological and ecological variations among species, and to a lesser extent, populations (Schwenk & Throckmorton, 1989; Butler,

2005; Measey et al., 2011).

While many predatory lizards may be easily classified into one of these two feeding strategies, Perry (1999) argued that they may not be mutually exclusive.

Chameleon feeding strategies are not as easily classified as their physiology, biochemistry, and muscular distribution limit locomotion (Abu-Ghalyun et al., 1988;

Butler, 2005). Unlike many actively foraging predators, chameleons do not typically wander in search of their prey; nor do chameleons always sit and wait for prey to land in front of them (Butler, 2005). In an absence of prey, chameleons will move short distances 32 to scan for prey. This intermediate foraging strategy has since been termed cruise foraging and has been used to describe feeding behavior in many species of chameleon

(Regal, 1978; Butler, 2005; Hagey et al., 2010; Measey et al., 2011). For the purposes of classifying chameleon feeding ecology, diet and foraging period seem to be the most important factors as foraging space and foraging-group size are not usually included in analysis (Anderson & Karasov, 1981; Butler, 2005; Measey et al., 2011). The unusual nature by which chameleons feed is just one reason that makes chameleon-feeding ecology difficult to study.

Since chameleons are constrained physiologically from moving quickly to pursue their prey, the percentage of time spent moving and rate of travel are valuable indications of feeding strategies (Abu-Ghalyun et al., 1988; Butler, 2005; Cooper et al., 2005;

Cooper, 2007). Radio telemetry, the practice of using radio waves to track organisms, offers freedom for the subject to move freely throughout the environment, while allowing the researcher to leave and return to the study site, but may be cost-prohibitive. Using cotton spools and line, the practice of physically affixing cotton spools to the organism and allowing the line to be drawn out, has become a cost-effective alternative to telemetry in reptiles and amphibians and has subsequently been used in several short- term movement and retreat studies (Tozeti & Toledo, 2005; Lemckert & Brassil, 2000;

Seebacher & Alford, 1999). Another method used to estimate distance traveled is through direct observational studies beginning and ending when chameleons sleep

(Altmann, 1973). Each method has advantages. Direct observations may occasionally interrupt the normal behaviors of the animal if the subject is aware of the observer, but can be remedied by the spooling technique. However with spooling, the observer 33 encounters another problem in that it is impossible to tell how far the animal actually traveled in a given time period since it is possible that the organism doubled back, thereby giving an inaccurate reading of the total distance traveled. Direct observations help to overcome these inaccuracies (Altmann, 1973).

Little is known of the general ecology of the chameleons that inhabit Bioko

Island, Equatorial Guinea, let alone their feeding ecology (Larison et al., 1999). Four chameleon species are known to live on Bioko Island, Trioceros feae, T. cristatus, T owenii, and Rampholeon spectrum (Larison et al., 1999; Necas, 1999; Barej et al., 2010).

This study focuses on two of the four chameleons found on Bioko, T. feae, a species endemic to Bioko and R. spectrum, a Western/Central African species also found in

Cameroon and Congo, (Necas, 1999). Both T. feae and R. spectrum can be found in the submontane forests of Moka. In contrast, T. cristatus and T. oweni are lowland species and are not known from the forests surrounding the village of Moka. Aside from its original description in 1906 and taxonomic classifications based on reproductive and lung morphology, little else is known of T. feae’s biology ( laver & B hme, 1992). Data on feeding strategies for R. spectrum or T. feae are lacking. Diet preferences also remain poorly studied for both species.

The goal of this study was to determine whether R. spectrum and T. feae utilized different feeding strategies imposing different energy requirements. In this study, I seek to categorize the feeding strategy of R. spectrum and T. feae as active foragers, cruise foragers, or sit-and-wait predators via direct observation of different individuals. Direct observations recorded data on color changes, intra- and inter-specific interactions, feeding attempts, and percentage of time traveled within 12 hours. An understanding or 34 the feeding strategies employed by the two chameleons is fundamental to understanding their energy requirements. In this study, the daily distance traveled, the daily percentage of time traveled, and the daily average rate of travel were observed to help to elucidate each species’ foraging strategy. It is hypothesized that as a larger species, T. feae is a more active forager than the smaller R. spectrum.

METHODS

Direct Observation

Data collection took place during the transition between the dry and wet seasons

(April 17 – June 8, 2012) in Moka, Bioko Island, Equatorial Guinea. A total of 22 adult chameleons (7 Rhampholeon spectrum males, 5 R. spectrum females, 5 Trioceros feae males, 5 T. feae females) were observed for behavioral data. Direct observation cycles consisted of recording feeding attempts, movement, inter- and intraspecific interactions, and color changes every 5 minutes for twelve hours. The observations began shortly after sunrise, along the Nature Trail (N 03˚21’41.3”, E008˚39’44.5”). Observers maintained a distance of at least 3 m, hiding behind cover wherever possible, and used binoculars whenever necessary so as not to disturb normal chameleon behavior. Color changes were used as a measure of observer disturbance as chameleons may turn black in color due to stress. No chameleon was resampled during the study. While left unmarked, the chameleons were identified by their location and unique natural markings. Upon completion of the 12-hour observation cycle, the total distance that individual chameleons traveled was measured to the nearest centimeter. 35

Observations were completed during days in which weather conditions did not prevent observations. Behavioral assessments that took place during conditions in the rainforest that were unsuitable or unsafe for observation (i.e. torrential rains, storms) were removed from the study. Since weather conditions may fluctuate throughout the day, causing additional variations in chameleon behavior, wind speed, wind direction, solar radiation, temperature, dew point and temperature were recorded by an Onset

HOBO weather station situated at the Moka Wildlife Center every fifteen minutes for the duration of each observation cycle. As the individual Bioko chameleons observed in this study did not ascend to the canopy and live in forested habitat, wind speed, wind direction, solar radiation, and dew point were regarded as irrelevant data for the purposes of this study and were not further examined as factors that would directly affect daily chameleon behavior.

Interspecific analysis of activity behavioral data

Five measures of foraging behavior were estimated to determine difference in energy/behavioral allocation to hunting strategy. Total distance traveled (TDT) was a direct measurement of the distance traveled, in centimeters, by each chameleon at the end of each observation. Average speed (AS) was calculated as the total distance traveled divided by the total time the chameleon was observed including the time when the chameleon was stationary, measured in cm/min. Movement speed (MS) was calculated as the total distance traveled divided by the total time the chameleon spent moving, measured in cm/min, total. Feeding attempts (FA), recognized by any event in which the chameleon’s tongue was projected, and percent of time spent moving (PTM) were also measured as a proxy for determining feeding strategy. Behavioral differences among 36 were tested to compare sexes within both species (adult male or adult female), as were associations between ambient temperatures, time of day, and activity level. Welch’s t- tests were performed to compare the means of TDT, AS, MS, PTM, and FA between sexes, given that their variances differed among all values. Behavioral data was treated as dependent variables; species was regarded as the independent variable. As sex was not significant in preliminary tests, both sexes were grouped to measure differences between species (Table 2, 3). Data are presented in means  SEM.

Effects of rain and ambient temperature on movement were analyzed using a binomial linear model (lmer) found in the lme4 package in R 2.15.2. A full model including species:sex interaction, species, sex, time, and temperature was first tested for their effects on activity level. Time was broken down into six 2-hour cycles. Maximum activity during one 2-hour cycle was given a score of 24, corresponding to twenty-four 5- minute bouts of movement. Subsequently, species:sex interaction, sex, species, and temperature were individually dropped and each model was retested. Variables that did not contribute significantly, reported as variables with lesser Akaike information criterion

(AIC), were dropped from further analysis. The predictive model was determined by the simplest model whose variables cannot be removed without significantly altering the estimated value of a chameleon’s activity level (Table 4).

Intraspecific analysis of activity behavioral data

The same measures of hunting behavior were used within species (T. feae or R. spectrum). Welch’s t-tests were used to compare the means of TDT, AS, MS, PTM, and

FA between species. Behavioral data was treated as dependent variables against the independent species variable. 37

RESULTS

A total of 22 individual chameleons (5 female R. spectrum, 7 male R. spectrum, 5 female T. feae, and 5 male T. feae) were observed over the course of the field season for

12 consecutive hours between 6:00 and 19:00. Ambient temperatures ranged from

13.7˚C – 23.5˚C, with an average of 17.9˚C  0.24 ˚C (SE). Two male R. spectrum observations were removed from analysis; one observation was removed due to heavy rainfall, while the other was removed from analysis because the observation did not last for more than 40% of the total expected duration of the observation. During an observation, R. spectrum individuals tended to descend to the ground frequently, while T. feae tended to move horizontally through its environment.

The total distance traveled (mean  SE) by male T. feae chameleons was 985 

246 cm a day as compared to the 679  196 cm traveled a day by female. R. spectrum males traveled 803  252 cm on average per day, whereas female R. spectrum traveled an average of 300  96 cm per day. The average distance traveled for both sexes during daily observations averaged 832  25 cm for T. feae and 551  216 cm for R. spectrum.

Distances traveled were not different between species or between sexes within species (T. feae sex: t= 0.972, df= 7.619, p=0.361; R. spectrum sex: t= 1.866, df= 5.139, p = 0.120; species: t= 1.282, df= 17.984 p= 0.216) (Fig. 8) (Table 2 & 3).

The average speed (reported as mean  SE) by male T. feae chameleons averaged

1.456  0.325 cm/min a day whereas females exhibited a mean average speed of 0.992 

0.255 cm/min. R. spectrum males exhibited a mean average speed of 1.192  0.327 cm/min while females exhibited a mean average speed of 0.455  0.121 cm/min. The combined mean average speed rate was 1.224  0.297 and 0.823  0.290 cm/min for T. 38 feae and R. spectrum, respectively. Average speeds were not significantly different between species and between sexes within species (T. feae sex: t= 1.124, df= 7.572, p=0.295; R. spectrum sex: t= 2.119, df= 5.069 p = 0.087; species: t= 1.365, df= 17.991, p= 0.189) (Fig. 9) (Table 2 & 3).

The mean moving speed for T. feae males was 3.109  0.718 cm/min whereas the female T. feae chameleons had a mean moving speed of 1.822 0.453 cm/min. Average moving speeds for R. spectrum males were 3.538  0.707 cm/min and 2.358  0.683 cm/min for females. Moving speed was 2.465  0.642 cm/min for T. feae and 2.948

0.712 cm/min for R. spectrum. Differences observed with respect to moving speed were not significant between species and between sexes within species (T. feae sex: t= 1.517, df= 6.748, p = 0.175; R. spectrum sex: t= 1.200, df=7.99, p=0.264; species: t= 0.712, df=

17.812, p= 0.189486) (Fig. 10) (Table 2 & 3).

The average percent of time spent moving during the observation cycles was

52.0%  0.06 for T. feae males and 54.1%  0.09 for T. feae females. R. spectrum chameleons showed a greater variance of average values. R. spectrum males spent 31.6%

 0.07 of their day moving whereas females only allocated 21.7%  0.2 of their day moving. Throughout the day, T. feae chameleons spend significantly more of their day moving at 53.1%  0.08 as compared to the 26.6%  0.07 of the day R. spectrum chameleons spend moving. Differences observed in percent of time spent traveling were not statistically significant in respect sexes within both species (T. feae sex: t= 0.512, df=

7.387, p = 0.624; R. spectrum sex: t= 0, df=7.639, p=1; species: t= -3.541; df= 16.933; p<0.01) (Fig. 11) (Table 2 & 3). 39

A multivariate analysis was performed on sexes of R. spectrum chameleons using

PTM, TDT, AS and MS as dependent variables (MANOVA) since t-tests approached the statistical significance threshold. The test failed to reject the null hypothesis (p > 0.05) due to a small sample size.

Feeding attempts were observed during 16 of the 20 chameleon observation sessions. The maximum number of daily attempts at feeding was 8 times within a particular observation while the minimum number of observed daily feeding attempts was 0; the average number of feeding attempts during a day was 2.8  1.2. On average, male T. feae chameleon attempted to feed 3.6  1.2 times during the day while females attempted to feed 4.6  1.6 times during the day, giving an average of 4.1  1.3 feeding attempts for T. feae. R. spectrum males and females both averaged 1.4  0.6 (m) and 

0.7(f) feeding attempts per day. Feeding attempts occurred more frequently in T. feae when compared to R. spectrum (t= -2.5967, df=12.981, p<0.05) (Fig. 12) (Table 2 & 3).

Captured prey were usually very small and unidentifiable to the naked eye except for one case in which a female T. feae fed upon a pentatomid (shield) bug in the insect order Hemiptera. Handling time of prey, the time in which the chameleon must devote to physically handling the prey that may limit its ability to eat additional items, was not observed to exceed 5 minutes except for the case of the pentatomid bug in which the handling time lasted 70 minutes.

Solar radiation, rain, and ambient temperature were plotted against activity level but most variables showed little interaction; the only variable that appears to have an effect on activity is ambient temperature. Rain and solar radiation data showed to no effects and were removed from analysis. Binomial regression models concerning 40 species:sex interactions, species, sex, time, and temperature were included in the full model to be first examined for their effects on activity level. Species:sex interactions and sex showed no correlation (p=0.3489; p=0.526, respectively). Temperature and species were both significant variables (p<0.005; p<0.001, respectively) (Table 4).

DISCUSSION

Analysis of foraging behavior showed similar behaviors between the two species.

Aside from number of feeding attempts and the percent of time spent moving, R. spectrum and T. feae had similar moving and average speeds. The average total distance traveled per day was less than 10 meters for both chameleon species. Average rates of movement during the most active fell within the range of 2−3 cm per minute but did not exceed 4 cm per minute. Sex, however, does not appear to be important in respect to activity and hunting behaviors, which may have resulted from the small sample sizes

(n=5). Visual assessment of the data appears to show differences in total distance traveled, moving speed, average speed, and the percent of time spent moving, but not in feeding attempts between males and females of R. spectrum. Likewise, visual assessment of T. feae behavioral data alludes to differences in MS between males and females, but not in other behaviors. In all cases of differential values, values for females were lesser when compared to males of the same species. This difference is most noticeable for R. spectrum females.

Sexual differences are supported by the binomial regression analysis on weather and its effects on activity level. Ambient temperature, the time of day, and species were important for determining activity level. The interaction between sex and species as well 41 as sex alone did not play a factor on the activity level of the chameleons. Ambient temperature and species were important and once removed from the analysis, greatly affected the resulting model. This may be due to the fact that variance between males and females in T. feae was not as extreme as variance between males and females in R. spectrum. It is expected to see greater activity levels with increasing ambient temperatures. As ectotherms, chameleon metabolism is driven in direct association by temperature. As their metabolisms increase, chameleons may need to increase their caloric intake, thereby inducing pressure to increase foraging activity.

While not significant in the study, male R. spectrum traveled greater average distances than females, suggesting that with more observations sexual differences in foraging and movement would be demonstrated. Territoriality in chameleons has been much debated. It was originally believed that chameleons were solitary and aggressively defensive against other chameleons (Necas, 1999). This has been examined showing extreme diversity in terms of chameleon territoriality. While the traditional view is true in some cases (i.e. Chamaeleo dilepsis), some species have been shown to live in dense populations and even form long-term pairs (Trioceros hoehnelii, Hebrard, 1983;

Toxopeus, 1988; Necas, 1999). Territoriality in males has often been described in respect to mate guarding and home ranges; however sedentary females have only recently been recorded (Cuadrado, 2000; Cuadrado, 2001; Butler, 2005). With greater samples sizes, if males are found to move larger distances than females, the data may suggest that R. spectrum females have established home ranges. The males may in fact circulate throughout the environment looking for suitable mates. Males and females of T. feae, on the other hand, did not exhibit similar behavior. Instead, a faster MS only means that 42 males exhibit a faster speed (cm/min) measured over the time they actually move.

However, statistical tests do not infer any difference in feeding strategy between the two sexes of T. feae.

There was no significant difference between species for any of the activity measures (Fig. 9 & 10). One method used to quickly and crudely discern an organism’s general feeding strategies in the active forager/sit-and-wait predator paradigm is to look at moving speed and average speed when comparing two species. Typically, more active foragers have a greater average speed, whereas sit-and-wait predators have a greater moving speed (Cooper, 2007). As the percent time spent moving increases for more active foragers, the distance traveled is divided by a greater number, thereby decreasing the value of moving speed. While the average speed and moving speed averages are greater and lesser respectively for T. feae when compared to R. spectrum (t = -1.3653, p =

0.189; t = 0.7122, p = 0.49), the differences are statistically not significant. It may be likely that a greater sample size would increase the statistical power of these tests, showing true differences in the average speed and moving speed means between species.

Trioceros feae and R. spectrum differ in the percent time spent moving and in the average number of feeding attempts per day. T. feae spends more time during the day moving with a daily average of 53.1% as opposed to R. spectrum’s 26.6%. This is characteristic of a more active forager (Butler, 2005; Cooper, 2007). Sit-and-wait predators do not actively peruse their environments in search of prey and will typically stay in one area until it locates prey close enough to pursue (Schoener, 1971). Feeding attempts were also significantly greater for T. feae. As some prey are difficult, it was difficult to determine whether capture success was near 100% as has been seen in 43 previous studies (Butler, 2005). Increased feeding attempts may be described in T. feae for a variety of reasons. In the event that feeding attempts were unsuccessful, chameleons would need to attempt to feed more to compensate for missed opportunities.

Likewise, if a chameleon expends more energy in foraging, as is expected for an active forager, it would need to feed in greater quantities or feed on higher-quality (more energy-dense) prey. One way to resolve this question would be to experimentally examine the energy expenditure of the chameleon as well as the energy density within prey using calorimetry and bomb-calorimetry, respectively.

Traditionally, organisms were split into two categorical feeding strategies: active foragers or sit-and-wait predators. This bimodal paradigm has since been reexamined as researchers realized that organisms operate along a continuum between the two described foraging strategies (Cooper, 2007). Methods previously used to determine feeding strategy only examined movements per minute and percent of time spent moving (Butler,

2005; Hagey et al., 2010). Cooper (2007) proposed examining a suite of 5 behavioral traits, using combinations of the traits to assess foraging strategy. In addition to moves per minute and the percent time spent moving, average speed was equally as important in resolving feeding strategy. Unfortunately, moves per minute was impossible to estimate from the methods utilized in this study, and as a result, it was difficult to confirm the foraging strategies employed by T. feae and R. spectrum. However, percent of time spent moving was significantly different between the two species, showing a greater PTM for

T. feae. While not statistically significant, MS could potentially be greater for R. spectrum and AS may be smaller in T. feae, if more observations were included, thereby increasing the sample size. These three measures would suggest that T. feae is a more 44 active forager that R. spectrum and would therefore spend more time and energy foraging.

Table 2. Intraspecific behavioral data: Total distance traveled (TDT), average speed (AS), moving speed (MS), percent time spent moving (PTM), and feeding attempts (FA) did not differ significantly with the change among sexes within species.

R. spectrum females R. spectrum males (n=5) (n=5) T P TDT 299.7  96.176 803.2  252.183 1.866 0.120 AS 0.455  0.121 1.192  0.327 2.119 0.087 MS 2.358  0.683 3.538  0.707 1.200 0.264 PTM 0.217  0.061 0.316  0.067 1.087 0.309 FA 1.4  0.748 1.4  0.600 0 1

T. feae females T. feae males (n= 5) (n=5)

TDT 678.9  196.2 985.0  246.299 0.972 0.361 AS 0.992  0.255 1.456  0.325 1.124 0.295 MS 1.822  0.453 3.109  0.718 1.517 0.175 PTM 0.541  0.088 0.520  0.061 0.170 0.869 FA 4.6  1.567 3.6  1.166 0.512 0.624

Table 3. Interspecific behavioral data: Total distance traveled (TDT), average speed (AS), and moving speed (MS) were insignificantly variable between species. Percent time spent moving (PTM) and feeding attempts (FA) differed significantly between species.

R. spectrum T. feae Total (n=10) (n=10) (n=20) T P TDT 551.45  215.546 831.95  691.7  222.464 1.282 0.216 221.984 AS 0.823  0.290 1.224  1.024  0.300 1.365 0.189 0.297 MS 2.948  0.712 2.465  2.707  0.669 0.712 0.486 0.642 PTM 0.266  0.065 0.531  0.399  0.095 3.541 0.00252* 0.083 FA 1.4  0.639 4.1  1.324 2.75  1.187 2.597 0.0222*

Table 4. Binomial models determining variables affecting activity levels: The full model included five variables (sex, species, time, temperature, and the interaction between species and sex). Subsequent models removed one sequential variable until significant changes occurred with the removal of species and time.

Model Variables Removed AIC p-value Variable Full Model spp*sex + timecd + ------temp + (1|ID) Reduced Model spp + sex + timecd + spp*sex 649.35 0.3489 temp + (1|ID) Predictive Model spp + timecd + temp + sex 647.75 0.526 (1|ID) Reduced Model timecd + temp + (1|ID) spp 656.60 0.0009899* Reduced Model spp + timecd + (1|ID) temp 654.00 0.004089*