CRITICAL THREATS to ASIAN ELEPHANT HABITAT in SRI LANKAN PROTECTED AREAS Christie Sampson Clemson University, Csampso@Clemson.Edu

Home , Catunaregam, Udawalawe National Park, Yala National Park

Clemson University TigerPrints

All Theses Theses

12-2013 CONSERVING THE LAND OF THE GIANTS: CRITICAL THREATS TO ASIAN ELEPHANT HABITAT IN SRI LANKAN PROTECTED AREAS Christie Sampson Clemson University, [email protected]

Follow this and additional works at: https://tigerprints.clemson.edu/all_theses Part of the Ecology and Evolutionary Biology Commons

Recommended Citation Sampson, Christie, "CONSERVING THE LAND OF THE GIANTS: CRITICAL THREATS TO ASIAN ELEPHANT HABITAT IN SRI LANKAN PROTECTED AREAS" (2013). All Theses. 1789. https://tigerprints.clemson.edu/all_theses/1789

This Thesis is brought to you for free and open access by the Theses at TigerPrints. It has been accepted for inclusion in All Theses by an authorized administrator of TigerPrints. For more information, please contact [email protected].

CONSERVING THE LAND OF THE GIANTS: CRITICAL THREATS TO ASIAN ELEPHANT HABITAT IN SRI LANKAN PROTECTED AREAS

A Thesis Presented to the Graduate School of Clemson University

In Partial Fulfillment of the Requirements for the Degree Master of Science Biological Sciences

By Christie Lynn Sampson December 2013

Accepted by: Dr. David Tonkyn, Committee Chair Dr. Saara DeWalt Dr. Peter Leimgruber

i ABSTRACT

Grasslands habitats are hypothesized to be a critical resource for the endangered

Asian elephant (Elephas maximus) throughout its range. However, this hypothesis has not been rigorously tested. My study examined elephant habitat selection in Sri Lanka to determine the importance of grassland versus other local habitats, and how livestock abundance, fire, and the invasive plant Lantana camara affect the relative abundance of elephants within habitats. My research was conducted in two protected areas in Sri

Lanka, Udawalawe National Park (UNP) and Hurulu Forest Reserve (HR). I used distance- sampling on a total of 50.8 km of dung line-transects in the four habitat types and under burnt and unburnt conditions to assess relative abundance of elephant and livestock populations. I also established 197 permanent vegetation plots in UNP and HR to determine plant composition and the extent of L. camara invasion. I found that relative elephant abundance is highest in grasslands, specifically areas with Panicum maximum. I found no significant difference in the abundance of elephants in burnt vs. unburnt habitats; however, livestock abundance is greater in recently burnt areas. Livestock and elephant populations are often found in the same habitat, indicating they may be competing for food resources. I conclude that grasslands are essential habitat for Asian elephants in Sri Lanka and that stopping illegal grazing and the spread of L. camara in protected areas should be considered vital components of future conservation efforts.

ii DEDICATION

I dedicate this manuscript to my family.

ACKNOWLEDGMENTS

First, I would like to thank my advisor Dr. David Tonkyn for all of his support over the past three years. And my graduate career would not have been possible without Dr.

Peter Leimgruber who gave me the opportunity to work on this project. Thank you to

Dr. Saara DeWalt for her analysis help and writing critiques. I am indebted to Dr. Pruthu

Fernando, Dr. Jenny Pastorini and the entire Centre for Conservation and Research team for all of their help and guidance. Thank you to my field crew, Nishantha, Sanka, and especially H.G. Janaka, who worked tirelessly with me in Sri Lanka and the interns at the

Smithsonian GIS lab for all of their assistance. The study was made possible with funding from the Smithsonian Institution and the United States Fish and Wildlife Asian Elephant

Conservation Fund (# F11AP00263). And finally, a big thank you to my friends and family.

iii TABLE OF CONTENTS

Page TITLE PAGE ...... i

ABSTRACT ...... ii

DEDICATION ...... iii

ACKNOWLEDGMENTS ...... iii

LIST OF TABLES ...... vi

LIST OF FIGURES ...... vii

CHAPTER

I. BACKGROUND ...... 1

II. CRITICAL THREATS TO ASIAN ELEPHANT HABITAT IN SRI LANKAN PROTECTED AREAS ...... 5

Introduction ...... 5

Methods ...... 9 Study sites ...... 9 Elephant and livestock relative abundance survey design ...... 10 Vegetation survey design ...... 14 Vegetation plot sampling ...... 14 Intensive vegetation sampling ...... 15 Timing of surveys ...... 15

Data Analysis ...... 16 Elephant and livestock relative abundance ...... 16 Vegetation ...... 17 Models ...... 18

Spatial Analysis ...... 18 Fire ...... 20 Models ...... 21

iv Table of Contents (Continued)

Results ...... 22 Dung decay ...... 22 Elephant and livestock relative abundance ...... 22 Plant community ...... 23 Fire response of plant species ...... 27 PCA ...... 29 Models ...... 29

Discussion...... 33

Conclusion ...... 39

Appendices ...... 41

1: Study sites ...... 42 2: Habitat type examples ...... 43 3: Detailed survey information ...... 44 4: Vegetation plot ...... 45 5: Plant species ...... 46 6: Habitat classification ...... 47 7: Correlation matrices ...... 49 8: PCA plot of Udawalawe National Park ...... 53 9: PCA plot of Hurulu Forest Reserve ...... 54

References ...... 55

III. LANTANA camara: ALTERING FIRE IN AN ECOSYSTEM ...... 61

Exotic plants around the world ...... 61

Lantana camara ...... 62 Origin ...... 62 Lantana camara and fire ...... 63

Post-fire regeneration of Lantana camara ...... 65 Seed regeneration ...... 66 Vegetative regeneration ...... 67 Alleopathy ...... 68

v Table of Contents (Continued)

Lantana camara and development ...... 69

Herbivores and Lantana camara ...... 70

Management options...... 72

Summary ...... 73

Literature cited...... 74

LIST OF TABLES

Table Page

2.1 Summary of line transect surveys ...... 12

2.2 Dung decay results ...... 23

2.3 L. camara invasion results ...... 27

2.4 PCA results ...... 30

2.5 Udawalawe National Park model results ...... 31

2.6 Hurulu Forest Reserve model results ...... 32

vii LIST OF FIGURES

Figure Page

2.1 Dung transect and vegetation plot survey locations ...... 13

2.2 Fire locations within Hurulu Forest Reserve ...... 21

2.3 Elephant and livestock abundance in four habitat types in Udawalwe National Park ...... 24

2.4 Elephant and livestock abundance in two habitat types in Hurulu Forest Reserve ...... 25

2.5 Elephant and livestock abundance in burnt and unburnt habitats in Hurulu Forest Reserve ...... 26

2.6 Changes in plant composition following a burn in Hurulu Forest Reserve ...... 28

3.1 Lantana camara fire cycle ...... 64

3.2 Community structure changes made by Lantana camara ...... 65

viii CHAPTER ONE

BACKGROUND

The Asian elephant (Elephas maximus) range has been reduced from approximately 2.87 million km2 in the early 1900s to approximately 620,000 km2 fragmented over 13 countries in the early 2000s (Fernando & Leimgruber 2011). Asian elephants are listed as endangered on the IUCN Red List (Choudhury et al. 2008) and have an estimated population of 30,000-50,000 (Santiapillai & Jackson 1990, Shoshani &

Eisenberg 1982, Sukumar 1992). However, this figure is based on expert guesses and not systematic surveys and is not believed to be a reliable estimate (Blake & Hedges 2004).

Despite recent efforts to improve population estimates (de Silva et al. 2011), adequate global information is currently unavailable.

Habitat loss and human elephant conflict (HEC) pose the greatest threats to the survival of the Asian elephant (Fernando & Leimgruber 2011, Fernando et al. 2005,

Sukumar 1989, Sukumar 2003). As elephant habitat is increasingly developed for human use, HEC rates rise, with detrimental consequences for both elephant and human populations (Flint 1994, Koh & Wilcove 2008, Leimgruber et al. 2003, Sodhi et al. 2004,

Uryu et al. 2008). These activities can lead to the local extinctions of elephant populations and broad-scale population declines. Protected areas such as national parks and reserves are providing some of the last refuges for Asian elephants in the wild.

1 Sri Lanka has only 2% of the elephant’s potential range (Perera 2009), yet is home to one of the largest remaining wild Asian elephant populations, supporting an estimated 3,500-4,000 individuals (Choudhury et al. 2008, Perera 2009). Owing to mass hunting and land clearing, Sri Lanka’s elephant population dropped from as high as

8,000-12,000 elephants before the 19th century (Fernando 2000, McKay 1973) to an estimated 1,500 individuals by 1951 (Norris 1959). This population has recovered, in part due to the unique land use practices and agricultural system within the country

(Fernando et al. 2005).

With few natural permanent fresh water sources, the Sri Lankan people have developed an irrigation system including hundreds of man-made tanks to store water year-round, providing a permanent water supply to the elephant population. These tanks also support a riparian plant community when maintained or become inundated with lotus and grass species when abandoned, both of which provide an excellent food source for elephants (McKay 1973, Fernando et al. 2005). Also, chena, a traditional slash and burn agriculture practice where small forest patches are cut and cultivated for one to two seasons before they are abandoned, has provided abundant food sources for elephants (Wikramanayake et al. 2004). This practice limited interaction between humans and elephants, and allowed the two species to ‘share’, minimizing HEC

(Fernando et al. 2006). As the country transitions to a more intensive and permanent agriculture industry, the sharing of fertile land has decreased and conflict between the species has risen (Fernando et al. 2005).

2 Sri Lanka has a very high human population, currently over 21 million people with a density of 324 people per km2 (Central Bank of Sri Lanka 2012). To satisfy its growing economic and health needs, Sri Lanka is continuing to expand its agriculture and invest in development. Agricultural contributions to the GDP grew by 5.2% between 2000 and 2010, and the Sri Lankan government allocated Rs.122.7 billion

(approximately 935.6 million USD) to road maintenance and development in 2012 for better transportation routes (Central Bank of Sri Lanka 2012). These demands for space and resources produce one of the highest rates of HEC within the elephant range, with

1,369 elephant deaths reported between 1999 and 2009, and a recent average of 150 elephant and 50-70 human deaths per year (Perera 2009). As development and habitat fragmentation proceed, elephant populations will decline and remaining populations will be pushed into the existing parks and reserves (Fernando 1997). Approximately

14% of Sri Lanka’s 65,600 km2 land mass is currently protected for conservation (Perera

2009). Despite this, a majority of the elephant population lives outside of conservation areas (Sukumar 1989) and pushing them into these protected zones may exceed their carrying capacities (Fernando et al. 2005, Fernando et al. 2006, Weerakoon et al. 2004), endangering other protected species as well. Scientists are just now conducting systematic assessments of how much elephant habitat is protected, and which of the currently protected habitats best supports increasing elephant populations (Fernando, pers. com.).

3 My study investigates which factors influence relative elephant abundance within habitats in two protected areas in Sri Lanka, Udawalawe National Park (UNP) and

Hurulu Forest Reserve (HR), and suggests management options that can help conservationists to identify and protect the most critical areas within the elephant range. This study is discussed in chapter two of this thesis. Chapter three contains a literature review of the effects of the invasive species L. camara on ecosystems.

4 CHAPTER TWO

CRITICAL THREATS TO ASIAN ELEPHANT HABITAT IN PROTECTED

AREAS WITHIN SRI LANKA

Introduction

Habitat loss and human elephant conflict (HEC) pose the greatest threats to the survival of the Asian elephant (Elephas maximus; Fernando & Leimgruber 2011, Fernando et al.

2005, Sukumar 2003, Sukumar 1989). As elephant habitat is increasingly developed for human use, human-elephant conflict (HEC) rates rise, with detrimental consequences for both elephant and human populations (Flint 1994, Leimgruber et al. 2003, Sodhi et al. 2004, Koh & Wilcove 2008, Uryu et al. 2008). These activities can lead to the local extinctions of elephant populations and broad-scale population declines. Protected areas such as national parks and reserves are providing some of the last refuges for

Asian elephants in the wild. We know little, however, about habitat use in these protected areas and how the presence of invasive species, livestock, or fire influences elephant habitat use.

Sri Lanka has only 2% of the elephant’s potential range (Perera 2009), yet is home to one of the largest remaining wild Asian elephant populations, supporting an estimated 3,500-4,000 individuals (Choudhury et al. 2008, Perera 2009). As agricultural development and habitat fragmentation escalate within Sri Lanka, elephant populations will decline and remaining populations will be pushed into the existing parks and

5 reserves. Approximately 14% of Sri Lanka’s 65,600 km2 land mass is currently protected for conservation (Perera 2009). However, a majority of the elephant population lives outside of these conservation areas (Sukumar 1989) and pushing elephants into these protected areas may exceed their carrying capacities (Fernando et al. 2005, Weerakoon et al. 2004), endangering other protected species as well. Currently, there are no systematic assessments of how much elephant habitat is protected and which of the currently protected habitats best supports increasing elephant populations.

Large grasslands are restricted to a few protected areas in Sri Lanka, where they were artificially created over the past decades, around reservoirs or by clear cutting timber. For example, forest cover in Udawalawe National Park (UNP) dropped from 267 to 28 km2 between 1956 and 1982, a loss of almost 90%, (Molegoda 1984). During this time, grassland and open areas increased from 5 to 150 km2. Previous research in other parts of the elephant range (Sukumar 1989, Sukumar 2003) suggests that grassland ecosystems may play a critical role in supporting elephant populations by providing abundant food resources to supplement forage lost to habitat fragmentation. Creating similar grasslands across Sri Lanka’s protected areas may significantly increase the carrying capacity for the species throughout the protected area network. Grassland ecosystems in Sri Lanka are not static (Molegoda 1984), and recent declines in this habitat are most likely due to fire suppression, resulting in scrub and forest encroachment and the spread of the invasive plant species Lantana camara.

6 Fire is widely used throughout Asia, the Americas, and Australia to maintain grasslands and prevent shrubs and trees from encroaching (Gadgil & Meher-Homji 1985,

Schmerbeck & Seeland 2007,van Wilgen et al. 2004). There is little natural fire in Sri

Lanka; most of the fires are started by humans, either accidentally or purposely to flush game or clear areas for agriculture and livestock (McKay 1973). Accounts from the 1970s have documented vast fires in Sri Lanka, consuming up to 75% of savanna ecosystems

(McKay 1973). Continued fire suppression may result in a loss of all significant grasslands within Sri Lanka.

Lantana camara (henceforth referred to as lantana), one of the world’s top 100 invaders (Lowe et al. 2004), can severely alter the structure (Duggin and Gentle 1998), composition (Gooden et al. 2009), and function of a landscape (Vitousek 1987), and change its fire regime (Hiremath & Sundaram 2005). This plant is prolific throughout

Africa, Asia, and Australia and is poised to invade some of the most sensitive and critical environments throughout many Asian semi-tropical and tropical habitats (Sharma et. al.

2005, Waterhouse & Norris 1987). Although elephants will use areas dense with lantana, they do not consume it (Wilson et al. 2013). Grass cover is reduced by the spread of lantana, directly diminishing the amount of available forage to sustain elephant populations. Habitat loss from invasive species has become an important conservation focus for Asian elephant habitat management (Wilson et al. 2013).

Habitat lost to either invasive species or to agricultural expansion not only threatens wildlife but also puts tremendous pressure on cattlemen to provide food for

7 their livestock. Cattlemen in Sri Lanka rarely own their own pastures, instead grazing their livestock on government lands where elephants also feed. Excluding them from protected areas to support endangered species often fosters resentment towards conservation (Mishra 1982). Illegal grazing of domesticated water buffalo (Bubalus bubalis) and cattle (Bos taurus) is rampant within Sri Lankan protected areas (pers. obs.) where, along with elephants, they are the major herbivores. Other herbivores in the protected areas including include sambar (Rusa unicolor), spotted deer (Axis axis), and barking deer (Muntiacus vaginalis). However, domestic herds are more abundant and reduce the forage available for elephant populations (Odadi et al. 2011). They can also alter the vegetation structure (Schulz & Leininger 1990) and diversity within an ecosystem (Szaro 1989), possibly posing an additional threat to elephants.

My study investigates which factors influence relative elephant abundance across different habitats in two protected areas in Sri Lanka, Udawalawe National Park

(UNP) and Hurulu Forest Reserve (HR). I use indirect population estimates derived from dung transects as well as detailed vegetation surveys and landscape-level land cover maps in UNP and HR to determine which habitats are used predominantly by elephants, whether fire affects elephant habitat use, and whether lantana and illegal grazing of domesticated animals are associated with lower elephant abundance within habitats. I also compare landscape-level and local-scale data sets to determine if habitat classifications from satellite imagery predict relative elephant abundance and vegetation classifications derived from field data. The results from this project will

8 provide crucial information to develop habitat management recommendations for elephant conservation in Sri Lanka and throughout the elephant range.

Methods

Study sites

I conducted surveys in two protected areas, UNP and HR (Appendix 1), which contain the only two large grassland habitats in Sri Lanka. The study areas are characterized by an average annual temperature of 28º C and annual rainfall of approximately 1,500 mm.

Field studies were conducted between June 2011 and August 2012 in UNP and between

September 2011 and August 2012 in HR. Both parks experience a bimodal distribution of rainfall (Zubair et al. 2008), with monsoons occurring from mid-October to December and March to May. Lantana is present in both UNP and HR, and both areas are subject to extensive illegal grazing by domesticated herds of cattle and water buffalo.

UNP was established in June 1972 with an area of approximately 30,000 ha in an area of former teak (Tectona grandis) and eucalyptus (Eucalyptus camaldulensis) plantations. UNP was created primarily to protect the catchment area of

Udawalawe reservoir, a man-made lake designed to provide water for agriculture and hydropower generation and managed by the Sri Lankan Department of Wildlife

Conservation (DWC). The park provides a refuge for a substantial Asian elephant population of 804 – 1,160 individuals (Silva et al. 2011), in addition to many other species of wildlife. The park is surrounded by a fence, but has two small corridors in the north allowing elephant migration in and out of the park. UNP is dominated by a large

9 grassland area east of the reservoir that transitions into scrub and secondary forest toward the northern and eastern borders of the park. It is the one of the only places in

Sri Lanka to study elephant habitat preferences across four habitat types: grassland, scrub savanna, scrub, and forest (Appendix 2). These habitat types were distinguished visually on the ground by vegetation composition and structure.

HR, also a former teak plantation, was designated as a bioreserve in January

1977 and is managed by the Sri Lankan Forest Department (FD). This approximately

25,000 ha reserve is composed primarily of dry evergreen forest with few permanent water sources and a substantial grassland located in the south of the reserve. This unfenced forest reserve is located next to several other protected areas, allowing for seasonal elephant migrations with large herds using the reserve during the rainy season when grasses are abundant.

Elephant and livestock relative abundance survey design

To quantify the relative abundance and distribution of the major herbivores within UNP and HR, I followed well-established protocols for indirect line-transect surveys (Barnes and Jensen 1998; Buckland et al. 2001; Hedges et al. 2005). I conducted over 50 km of dung transect surveys during the two field seasons, and quantified elephant and livestock relative abundance using distance sampling. In UNP, I established 23 1-km randomly orientated transects (Figure 1), stratified across habitat types: grassland, scrub savanna, scrub, and forest (Appendix 3). In HR, four major fires occurred in HR grassland and scrub savanna habitats in 2011, ranging from 5 to 30 ha in size. There I created 5

10 post-fire 400 m – 500 m transects, each with a duplicate, unburnt control transect located in a similar habitat immediately outside of the burn area.

I located dung piles along each transect, recording the perpendicular distance in centimeters from the center of each dung pile to the transect line, and assigned the approximate age of each sample as either <1 day, 1-4 days, 5-10 days, or more than 10 days based on appearance and moisture content. I assumed that each dung detection was an independent event, defecation rates were unaffected by habitat type, and dung piles were independent of the transect location (Thomas et al. 2010). Because of the difficulty in distinguishing cattle dung from water buffalo dung, I recorded the dung from both species as “livestock”.

In conjunction with the line-transect surveys, I conducted dung decay studies in

UNP to determine if there was a difference in the decay rate of the elephant and livestock dung between habitat types. Beginning in July 2011, I located 85 fresh (<24 hr old) elephant dung samples in four habitats: grassland (n=39), scrub (n=10), scrub savanna (n=12) and forest (n=24). I also monitored 14 livestock dung samples during the same time period in three habitat types: grassland (n=5), scrub (n=3), scrub savanna

(n=6), but was unable to locate fresh livestock dung in the forest habitat. I marked and numbered each dung sample with flagging and recorded its location with a GPS unit. I revisited each sample at 2 to 6 week intervals to monitor decay rate over time until the sample had completely decayed beyond recognition as a dung sample.

11 Table 1. A summary of line-transect dung surveys to determine relative abundance of herbivores at Udawalawe National Park (UNP) and Hurulu Forest Reserve (HR). In UNP, I surveyed 5875 m of grassland, 8795 m of scrub savanna, 3680 m of scrub, and 4550 m of forest, during each survey period. In HR, I surveyed 1900 m of burnt grassland and 1500 m of control grassland habitat, and 500 m of burnt scrub savanna and 900 m of control scrub savanna, each survey period.

Protected Survey Survey Dates Season Survey Variables* Area Name UNP U1 June 2011-October 2011 Dry Habitat type (G,T,S,F) UNP U2 July 2012 - August 2012 Dry Habitat type (G,T,S,F) HR H1 September-October 2011 Dry Habitat type (G,T), fire HR H2 December 2011 Wet Habitat type (G,T), fire HR H3 February 2012 Dry Habitat type (G,T), fire HR H4 April 2012 Wet Habitat type (G,T), fire HR H5 May 2012 Wet Habitat type (G,T), fire HR H6 August 2012 Dry Habitat type (G,T), fire *G=grassland, T= scrub savanna, S=scrub, and F= forest, and fire = recently (< 1 year) burnt area

Figure 1. Location of dung transects and vegetation plots within study areas at Udawalawe

National Park (UNP) and Hurulu Forest Reserve (HR).

13 Vegetation survey design

I conducted vegetation surveys to assess plant community composition, including the availability of elephant forage and presence of lantana. I established 139 vegetation plots in UNP and 58 in HR and sampled the vegetation using a point-intercept method.

At UNP, I used these surveys to analyze differences in plant composition and the extent of lantana among habitat types. In HR, I used the vegetation surveys to compare plant community composition between recently burnt (within one year) and unburnt areas and to monitor changes in plant composition following a burn.

I established all plots along dung survey transect lines, spaced 200 m apart along the habitat comparison transects in UNP and 100 m apart along the burn and control transects in HR. I marked the center of each plot with a PVC pipe, which was used, along with GPS coordinates to relocate the plots during subsequent surveys. Unfortunately, many of the pipes were trampled or removed by wildlife or people, making it difficult to locate the exact center of the plot. The GPS coordinates and local landmarks were used to get the closest possible (within 9 m) approximate origin.

Vegetation plot sampling

I used a small diameter aluminum pipe marked at 0.5 m, 1 m, 1.5 m, and 2 m and held vertically to sample vegetation at each height interval. I sampled the vegetation every 1 m along each of the four perpendicular 10 m long arms of the vegetation plot (Appendix

4). Any vegetation that the pipe intersected along the edge closest to the tape was recorded by species and height. I identified to species two common and invasive grasses

14 favored by elephants, Panicum maximum and Cogon imperata (Appendix 5). All other grasses were assigned to two categories: medium grasses were ≥ 25 cm in height, and short grasses were < 25 cm. I classified all shrubs and trees by species, along with the common herbaceous plants. I also documented bare ground, wood, water, rock, and duff as indicators of microhabitat conditions.

Intensive vegetation sampling

I conducted a 5 m x 5 m woody plant species inventory within the top left quadrant of every plot. I counted the number of each species of shrubs and trees above 0.5 m tall and assigned height classifications: 0.5- 1 m, 1 – 1.5 m, 1.5 – 2 m, and > 2 m. Because elephants tend to break trees, resulting in a high density of coppiced trees in the landscape, tree size was measured as diameter at the ground height (dgh) of any tree with a dgh ≥20 cm. If a plant had multiple stems from a common root at the ground level, I recorded each one separately.

Timing of surveys

In UNP, I completed two rounds of dung and vegetation surveys, between July and October 2011, and between June and August 2012 (Table 1). In HR, I conducted dung and vegetation surveys approximately every two months for a total of six rounds between September 2011 and August 2012.

15 Data Analysis

Elephant and livestock relative abundance

I used the software program DISTANCE 6.0 (Thomas et al. 2010) to indirectly assess relative elephant and livestock abundance from dung counts observed along transect lines across all major habitat types at UNP and HR, including burnt and unburnt areas at

HR. DISTANCE requires five components (Thomas et al. 2010): (1) survey data, (2) a data filter, (3) a model definition, (4) dung decay rates, and (5) defecation rates.

The dependent variables in the survey data entered into DISTANCE were the elephant and livestock dung counts and the perpendicular distance from the transect line to each dung pile. The data filter truncated the data to include only the 95% of dung found closest to the transect line to improve data fit (Buckland et al. 2001). I utilized the half-normal key for the dung detection, to best fit the data within the model. Dung decays rates were determined from the in-situ field study at UNP, and defecation rates were assumed to be constant across habitat types and within the burnt area (Fernando pers. com.). A Kruskal-Wallis rank sum test was used to compare differences in decay rates between the habitat types for each of the two animal groups. In HR, I compared relative elephant and livestock abundance solely during the wet season (surveys H2, H4 and H5 in Table 1) because elephants were not present in the area during the dry season. I selected the best model for both the elephant and livestock dung distribution from the line transect data based on the Akaike Information Criterion (AIC).

16 Vegetation

I used the point intercept data collected from the vegetation plots in two ways: to create a local dataset for each vegetation plot in UNP and HR and to monitor habitat recovery in HR following a burn. To create the plot-level dataset, I tabulated the number of occurrences of each environmental characteristic (bare ground, wood, water, rock or duff) and plant species, regardless of the height level in which it was recorded, across the 40 point intercept locations in each plot at UNP and HR. This number gave me a percent cover estimate for each environmental characteristic or plant species. I incorporated these data, along with woody plant and common herbaceous species counts, into a dataset that defined local plant and environmental conditions for each plot. To determine the percent cover per habitat, I separated the plots by the habitat type assigned during the initial survey stratification, and averaged the values by plant species or environmental characteristic.

In HR, I monitored habitat recovery following a fire by classifying each plant species and environmental characteristic into the following groups: grasses, woody plants, herbaceous plants, and environmental characteristics (bare ground, exposed rock, dead and down wood, and duff). These groups represent foods that the elephants graze (grasses), foods that they browse (woody plants), plants that may or may not be eaten (herbaceous plants), and the changes to the ecosystem that can follow a burn. I calculated the total number of point intercepts by plot to compare cover of each of the four groups between burnt and unburnt plots by habitat type.

17 Models

I developed competing statistical habitat models in order to assess the relative importance of environmental factors in determining relative elephant abundance across local and landscape scales. I used the number of dung piles found along each transect as an indicator of relative elephant abundance and the dependent variable. I assumed that the dung counts had a Poisson distribution and used generalized linear models to examine the relationship between this relative abundance indicator and different sets of continuous predictor variables. I created four models for each protected area, one based solely on landscape-scale data and three including local-scale variables derived from vegetation plot data.

Spatial Analysis

I used spatial analysis techniques to determine habitat qualities, including plant and livestock relative abundance, which can be used to predict elephant relative abundance.

All of the vegetation plots and dung transect lines were spatially referenced using a GPS unit. I entered these data and the coordinates for each individual dung pile into ESRI

ArcMAP 10.0 (ESRI 2011) and used them as the basis for the data analysis. For spatially- explicit models of elephant habitat selection, I divided transects into 50 m x 50 m grid cells centered on the transect line, and tabulated elephant and livestock dung counts and environmental covariates within each cell. I used the most recent habitat maps of

UNP and HR to provide landscape-scale landcover information (grassland, scrub, forest, agriculture, bare ground, floodplain, and water) (USFWS report 2013). These maps were

18 created from georeferenced, high-resolution satellite imagery provided by Google Earth, which was then analyzed in eCognition ©. Though I was able to define four habitat types when conducting the dung and vegetation surveys, this software could only distinguish three habitat types: grassland, scrub, and forest; therefore, the scrub savanna habitat type was not included in the landscape-scale data. A comparison between ground- truthed and satellite data revealed that in 54% of the scrub savanna habitat in UNP and

83% of the scrub savanna habitat in HR was classified as grassland by the software program. The remaining amount was classified as either scrub, forest, or other which was contained landcovers defined as floodplain, bare ground, building, or water. This information, including an accuracy assessment between the satellite and ground- truthed data, can be found in Appendix 6. The area of grassland, scrub and forest habitats was calculated for each cell.

I hand-digitized visible water sources from satellite imagery in Google Earth and calculated the mean distance from each cell to the nearest water body. I also digitized the boundaries of the four fires that occurred in HR by walking the boundaries with a handheld GPS unit. I used ArcMAP to spatially join the dung counts and the landscape- scale datasets: habitat types, the distance to the nearest water source, and the burn area. I then extracted the data contained with each analysis unit for statistical analysis.

Local-scale data were derived from vegetation plots along each of the transects. I conducted a principle components analysis (PCA) to transform the plant species and environmental characteristics data with the greatest percent cover recorded in the

19 point-intercept survey for grassland habitats into linearly uncorrelated values. This simplified the data, allowing me to test how variation in composition may affect habitat selection in elephants. In the PCA, I included the percent cover of lantana and the four types of grasses located within the plots: short grass, medium grass, P. maximum and C. imperata. I also included the percent cover of teak in the PCA for UNP because it is a very common species in the grassland there. In the HR PCA analyses, I included percent bare ground because of the high frequency of this cover type in the burnt areas. Cells that included vegetation plots were assigned the associated local-scale vegetation data derived from the PCA.

Fire

Only two small fires occurred in scrub savanna habitat in UNP covering a total of 23 ha in 2011.

However, four large fires occurred in HR (Figure 2) totaling 84.6 ha. Therefore, all of the results regarding fire are based on data collected at HR. I was only able to conduct a limited number of surveys at HR, so all of the transects were located in the most fire prone habitats, grassland and scrub savanna.

Models

The first models for UNP and HR, UNP-landscape and HR-landscape respectively, were based on landscape-scale data and tested if elephant use differed among the three habitat types that were identified by satellite images. For the local-scale models, I included the first two axis scores derived from the PCAs as independent variables in the

Figure 2. Locations of fires occurring between June and October 2011 in Hurulu Forest Reserve.

analysis instead of using the large suite of individual, correlated vegetation indices collected in the vegetation plots. Models UNP-local and HR-local included just these two axes. UNP-water and HR-burn included these axes as well as mean distance to water in UNP-water and percent burn area in HR-burn. For the final models (UNP- livestock and HR-livestock), I included the scores of the first two PCA axes and livestock dung counts, representing potential competition for resources. I also conducted a correlation analysis between the covariates for the landscape- and local-scale data to

21 identify any bias within the model results (Appendix 7). I conducted all statistical analyses in R version 3.0.1 (R Core Team, 2013).

Results

Dung Decay

Elephant dung decay rates did not differ significantly among the four habitat types

(Kruskal-Wallis chi-squared = 3.4, df=3, p= 0.33). Similarly, there was no significant difference (Kruskal-Wallis chi-squared = 4.5, df=2, p= 0.11) found for the livestock dung decay rates among the three habitat types (Table 2).

Elephant and Livestock Relative Abundance

Using the DISTANCE program and the 4 habitat types determined in situ, I found relative elephant abundance was significantly greater in grassland areas, followed by scrub, scrub savanna areas, and forest at UNP (Figure 3). Livestock in UNP used the grassland and scrub savanna areas most frequently, followed by scrub. In HR, there was no difference in relative elephant abundance between grassland and scrub savanna habitats within each round; however there was an increase in relative abundance from the H2 survey in February 2012 to the April and May 2012 surveys, H4 and H5 (Figure

4). During survey H2, elephant abundance was greater in the unburnt areas, but in subsequent surveys there was no difference between burnt and unburnt areas (Figure

5). The relative livestock abundance was higher in the grassland during surveys H2 and

22 Table 2. The number of dung decay samples for elephants and livestock by habitat with the range, mean and standard deviation (SD) in the number of days until the samples had completely decayed and the standard deviation of the time to decay .

Number of Number of days Habitat Species samples until decay Average SD Grassland Elephant 39 25-117 87 33 Scrub Savanna Elephant 12 71-144 73 24 Scrub Elephant 10 38-117 66 19 Forest Elephant 24 25-117 86 34 Grassland Livestock 5 23-72 37 3 Scrub Savanna Livestock 6 7-36 19 10 Scrub Livestock 3 11-22 18 5

H4, but there was no difference in abundance between grassland and scrub savanna habitats survey H5. However, relative livestock abundance was consistently higher in burnt than unburnt areas.

Plant communities

Panicum maximum and Tectona grandis were the most abundant plant species found in grassland habitats. Short grass and medium grass were dominant in scrub savanna habitats. Flueggea leucopyrus and Catunaregam spinosa were dominant in scrub habitats, and Croton laccifer and Psilanthus wightianus were the most abundant plants in forest habitats. Lantana was widespread in scrub savanna, scrub and forest-edge habitats, occurring in 52% of scrub savanna and scrub plots and 75% of forest plots,

23 Figure 3. Elephant and livestock relative abundance in Udawalawe National Park within the different habitats types: G=Grassland, S= Scrub, T = Savanna Scrub, and F=Forest, during the

2011 and 2012 survey periods. The graphs are scaled to reflect relative abundance so that for every one animal in the grassland habitat, a smaller proportion is found in the less- used habitat.

though the average cover per plot was only 1.59%, 3.65% and 2.78% respectively (Table

3). Lantana presence was much lower in HR, occurring in 18% of grassland and 44% of scrub savanna plots with an average cover of 1.74% and 0.77% respectively. While the average cover was low, individual plots in UNP and HR had lantana coverage up to 23% and 14.5%, respectively.

Figure 4. Elephant and livestock relative abundance in Hurulu Forest Reserve within the different habitats types, G = Grassland and T = Savanna Scrub, and during the separate survey periods, H2 = December 2011, H4 = April 2012, and H5 = May 2012. The graphs are scaled to reflect relative abundance so that for every one animal in the primary habitat, a smaller proportion is found in the less-used habitat.

Figure 5. Elephant and livestock relative abundance in Hurulu Forest Reserve within burn and control areas, B= burn and C= control, and during the separate survey periods, H2= December

2011, H4= April 2012, and H5 = May 2012. The graphs are scaled to reflect relative abundance so that for every one animal in the primary habitat, a smaller proportion is found in the lesser used habitat.

26 Table 3. Lantana presence by habitat types, expressed as number and percentage of plots with

Lantana as well as average cover and cover range at Udawalawe National (UNP) and Hurulu

Forest Reserve (HR). Average and range of percent cover were determined from the point intercept data collected in the vegetation plots.

Protected Average Cover Habitat* # of Plots Area Cover Range UNP G 32 (46%) 0.64% 0-3.5% UNP T 49 (52%) 1.59% 0-15.5% UNP S 48 (75%) 3.65% 0-23% UNP F 26 (52%) 2.78% 0-17.5% HR G 16 (44%) 1.24% 0-14.5% HR T 4 (18%) 0.77% 0-1% *G=Grassland, S= Scrub, T = Savanna Scrub, and F=Forest

Fire response of plant species in HR

In both grassland and scrub savanna habitats, the herbaceous and woody plant composition in burnt areas approached a composition similar to that of the control plots within 10 to 12 months (Figure 6). Environmental characteristics such as percent bare ground also approached control levels after 12 months. Grass species, however, never approached control composition levels in either grassland or scrub savanna habitats during the 12 months of study. Grass cover in burn plots only reached 26-63% of that in control plots within grassland habitats and 44-79% of control plots in scrub savanna habitats.

27 Oct-11 Dec-11 Feb-12 Apr-12 May-12 Aug-12 Survey Date

Oct-11 Dec-11 Feb-12 Apr-12 May-12 Aug-12 Survey Date

Figure 6. The effect of fire on the establishment of grasses, herbaceous and woody plants as well as environmental conditions (bare, water, rock and duff) over a one-year period. These data were based on total number of point intercepts counted per survey.

28 PCA

In UNP, the first two axes contained 83% of the variation (Appendix 8). The variables with the strongest loadings were percent cover of P. maximum and short grass along the first axis, and short grass, P. maximum, and medium grass along the second axis (Table

4). For HR, 81% of the variation was contained in the first two axes. Bare ground and P. maximum had the strongest loadings on the first axis, and P. maximum and short grass had the strongest on the second (Appendix 9).

Models

Models UNP-livestock (Table 5) and HR-local (Table 6) provided the lowest AIC scores with the fewest covariates within the respective study sites. Both of these models included the first two axes from the PCA of the vegetation data and UNP-livestock also included the livestock dung counts. However, all three of the HR models that included the local-scale data had similar AIC values.

29 Table 4. Principle components analysis loading from the first two axes for Udawalawe National

Park (UNP) and Hurulu Forest Reserve (HR). The * indicates the value is significantly different from zero at p< 0.05.

UNP HR Variable PC1 PC2 PC1 PC2 % Bare Ground - - -0.79* 0.38* % C. imperata 0.01 -0.03 -0.16* -0.01 % Lantana 0.07* 0.06* 0.01 -0.01 % Medium Grass 0.08* -0.37* -0.02 -0.10 % P. maximum -0.97* 0.17* 0.57* 0.70* % Short Grass 0.21* 0.91* 0.17* -0.60* % Teak -0.06 -0.04 - -

30 Table 5. Summary of the four models designed to predict elephant habitat use in Udawalawe National Park, with the covariates included in each model.

Model Name Model Variables Estimate Standard Error z value AIC ∆ AIC (Intercept) 1.65519 0.04802 34.469 *** 2339.4 UNP-livestock PCA1 0.00186 0.00130 1.429 PCA2 0.01528 0.00268 5.691 *** Livestock dung 0.03575 0.00364 9.815 *** * (Intercept) 1.73499 0.04669 37.163 *** 2412.4 UNP-local PCA1 0.00105 0.00127 0.827 PCA2 0.01734 0.00270 6.42 *** 73 (Intercept) 1.75400 0.04938 35.513 *** 2413.1 UNP-water PCA1 0.00128 0.00129 0.992 PCA2 0.01749 0.00270 6.48 *** MEANwater -0.00007 0.00006 -1.123 73.7 (Intercept) 2.48500 0.05949 41.764 *** 7283.3 UNP-landscape Grassland -0.00021 0.00003 -8.076 *** Scrub -0.00010 0.00003 -3.69 *** Forest -0.00101 0.00004 -24.073 *** 4943.9

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

31 Table 6. Summary of the four models designed to predict elephant habitat use in Hurulu Forest Reserve, with the covariates included in each model.

Model Name Model Variables Estimate Standard Error z value AIC ∆ AIC (Intercept) 1.53010 0.09429 16.228 *** 842.1 HR-local PCA1 0.01225 0.00384 3.188 ** PCA2 0.01411 0.00540 2.616 ** * (Intercept) 1.67600 0.14370 11.664 *** 842.33 HR-burn PCA1 0.01440 0.00416 3.463 *** PCA2 0.01769 0.00602 2.939 ** BurnArea -0.00006 0.00004 -1.327 0.23 (Intercept) 1.57915 0.10235 15.429 *** 842.71 HR-livestock PCA1 0.01347 0.00397 3.394 *** PCA2 0.01496 0.00540 2.771 ** Livestock dung -0.00663 0.00586 -1.132 0.61 (Intercept) 1.21900 0.19290 6.321 *** 1370.3 HR-landscape Grassland 0.00001 0.00008 0.064 Scrub -0.00015 0.00012 -1.319 BurnArea 0.00001 0.00003 0.397 527.59

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

32 Discussion

The DISTANCE results from the dung transect surveys indicate that elephants favored grassland habitats over the other three types at both HR and UNP, although at HR they also used scrub savanna habitats. This outcome mirrors observations made by McKay

(1973) over forty years ago in another Sri Lankan protected area, Gal Oya, with similar habitat types. Although elephants eat a range of plant types, grasses and sedges constitute an important part of their diet (McKay 1973, Sukumar 1990) and these plants are predominant in the grassland habitat. Livestock were also most abundant in grassland habitats in the two protected areas. In HR, elephants used both burnt and unburnt areas, however livestock favored burnt areas.

Dung counts are often the most practical survey method for estimating elephant population sizes and provide similar results as other procedures, such as aerial surveys, direct observation, and camera trapping (Barnes 2001). There are, however, several problems with this method, generally arising from the estimations of defecation and decay rates, rather than the transect survey themselves. While my results showed no difference in the decay rates between habitats, other studies have shown both defecation and decay rates can vary with habitat and season (Barnes 1982, Barnes et al.

1997, Guy 1975, White et al. 1995). My dung decay studies were conducted during the dry season and therefore could not be used to estimate decay rate for the surveys conducted in HR during the wet season. I chose not to use dung decay rates from other studies because of potential differences in microclimate variables such as insect

33 presence, fungi and plant germination, and environmental conditions such as exposure to sun (Pastorini et al. 2007) and rain (White et al. 1995; Barnes et al.1997; Nchanji &

Plumptre 2001), which can all alter decay rates and introduce error into the estimates.

Similarly, differences in vegetation consumption can alter defecation rates (Barnes

2001). Instead, my results are comparable within the time frame of each survey (i.e., all results from H1 are comparable to each other but not to H2, conducted two months later) and study site, but do not provide an accurate estimate of true elephant or livestock population within the protected areas or habitats.

I found little evidence of lantana invasion into the grassland in either UNP or HR.

However, I did find large patches within the neighboring scrub savanna and scrub habitats in UNP. Lantana was found in over half of all of the vegetation plots surveyed in the scrub savanna, scrub, and forest habitats in UNP, and there was a moderately positive correlation (0.24) between lantana and scrub habitats. Vegetation plots in HR had a much lower lantana presence in both the grassland and scrub savanna habitat.

The best supported models for relative elephant abundance in both HR and UNP included local-scale data. Elephants were found more often in areas with an abundance of short grass and with less P. maximum and medium grass, indicating that the presence of elephants is governed by the presence of short grasses. Livestock presence was also positively correlated with elephant presence in both locations, and its inclusion in the model was supported at HR. It does not appear that lantana has any influence on

34 elephant abundance at this local scale, but this result could be an aspect of the overall low percent cover of lantana within the vegetation plots.

The HR-local model included the first two axes of the PCA on which P. maximum was abundant, and bare ground and short grass were sparse. This indicates that elephants are most abundant in areas heavily populated by grasses including the invasive P. maximum during the wet season in HR. Bare ground was positively correlated with grassland areas in HR, most likely as a result of burns and overgrazing by livestock.

Observations by H.G. Janaka (pers. com.) of extensive livestock use of burnt areas at the start of the wet season, leading to significantly less grass cover in comparison to unburnt plots, were supported by my dung surveys and plant monitoring.

While elephants are habitat generalists (McKay 1973), it has been suggested that grassland and savanna habitats can support much higher Asian elephant densities than forests. This impression has been around for years, however little to no systematic studies have been conducted to test it. Most of Sri Lanka’s elephants currently survive in areas of dry zone vegetation dominated by dry evergreen forests, interspersed with chena, small-scale plantations, and artificial reservoirs. As this land use mosaic disappears is replaced with large scale intensive agriculture, habitat management will need to focus on managing remaining habitat to support the elephant populations and minimize HEC.

My results indicate that grasslands are providing a crucial foraging area for elephant populations in protected areas. However, grasslands do not naturally occur in

35 Sri Lanka, found only in the floodplains of man-made tanks and abandoned clearcut plantations (McKay 1973, Molegoda 1984). Savannas such as the Talawa in Gal Oya

National Park and those found throughout UNP are also rare and probably maintained through the frequent application of fire. Open grassland ecosystems are artificial, and P. maximum is an exotic, invasive species. Maintenance of this habitat will require intervention to prevent succession from grassland back to a scrub forest, by conducting prescribed burning or manually removing of trees and scrub along the edge habitats.

Both of the study sites are subject to active fire suppression administered by park and forest officials and military personnel. The largest fire I recorded during the two fire seasons in HR and UNP was just under 30 ha, which is relatively small in comparison to accounts of burns within the protected areas from the last century (McKay 1973). These fires also appeared to be of low intensity with a patchy burn pattern, leaving a mosaic of burnt and unburnt vegetation. Monitoring post-fire growth of lantana in scrub savanna habitats showed that the fires did not burn hot enough to kill lantana plants, and lantana was one of the first plants to begin resprouting following a burn (pers. obs.).

While lantana currently has a low average percent cover, lantana was found in 46% of the grassland vegetation plots and sometimes at high levels of percent cover, indicating a wide extent and potential for substantial increase in abundance if the habitat is disturbed.

Lantana creates a positive fire feedback loop similar to the grass-fire cycle (Hiremath

& Sundaram 2005). Once it has invaded an area, it stimulates higher fire frequency and

36 intensity, causing more disturbances and promoting invasion of more lantana, thus repeating the cycle (Hiremath & Sundaram 2005). Native vegetation may not be able to compete successfully for space and resources at this increased rate of disturbance. The majority of lantana within the forest habitats was found alongside the road and in abandoned chena plots. Lantana is a light-dependent species, unable to survive in the low light conditions of a mature forest with a dense, intact canopy (van Oosterhout et al. 2004). It favors highly disturbed areas, often found in abandoned agricultural fields, along roads, and at forest edges. Accidental fires generally occur near roads, adding an additional element into an already altered area, priming it for a lantana invasion.

Lantana was one of the first plants targeted for biological control, but little progress has been made at containing the species (Day et al. 2003, Julien & Griffiths

1998, Zalucki et al. 2007). Efforts for biological control began in 1902 but have not been successful despite the release of 41 agents in 40 different regions (Day et al. 2003, Julien

& Griffths 1998). Lantana is difficult to remove once established partially because it is capable of seed banking (Vivian-Smith & Panetta 2009) and layering (Day et al. 2003).

Given the extreme difficulty in removing an infestation once established and the devastating impacts this plant can have on the structure and composition of an ecosystem, managers of areas with the potential for lantana invasion should attempt to prevent any disturbance which could advance the spread the species. Sri Lanka, and most of the elephant range, however, is experiencing rapid growth, and conservation officials need to consider future management issues of the disturbed natural areas this

37 development is creating. Agriculture is fragmenting once-continuous forest, creating edge habitats and exposing forest to invasion. The need for lantana management is recognized within the Sri Lankan government and by the public (Ranwala & Thushari

2012). However, promoting public education and outreach would benefit the eradication of the species. For example, villagers have been able to reduce or even clear lantana from home sites through continuous removal methods, but they often dump the cut vegetation at the edge of protected areas, promoting the spread of the species into these areas (Ranwala & Thushari 2012).

The illegal grazing of livestock in the protected areas might also be furthering the spread of lantana. Gentle & Duggin (1997) examined the role of cattle in the promoting the growth of lantana in a dry rainforest ecotone in NSW, Australia. They determined that the biomass reduction and soil disturbance caused by cattle can increase lantana success. This relationship was primarily driven by grazing, which reduced the above ground biomass, increasing light penetration to the soil and any lantana seeds or seedlings.

Based on field observations from Sri Lanka, there may be another aspect of the livestock’s usage of the environment with an even greater effect on the spread of lantana. Throughout many areas in India and Sri Lanka, ranchers illegally graze their livestock on protected areas. The size of the domestic herds of livestock are far greater than the wild herds of water buffalo and other large bodied herbivores, such as sambar

(Rusa unicolor) and spotted deer (Axis axis). These domestic livestock create trails where

38 vegetation has been trampled down to bare earth, which are common in many of the protected areas in Sri Lanka. In slow burning, low intensity fires, these trails act as fire breaks, stopping or redirecting the spread of fire through the landscape and could assist in the spread of lantana.

Conclusions

Conservation areas have proven to be instrumental in the survival and recovery of

African elephants (Blanc et al. 2007) and they could play a similar role for Asian elephants. As Sri Lanka shifts from chena to permanent agricultural centers, protected areas will play a growing role for Asian elephants. Grasslands are important resources for the elephant populations, and managers need to actively maintain and promote grassland growth. In addition to improving elephant habitat, grasslands offer better wildlife viewing opportunities in an ecotourism-based economy. Conducting prescribed burns and removing trees and shrubs are two ways to expand grassland habitats.

Continued monitoring for lantana within the grassland needs to be done to ensure fire and disturbance due to removal of woody vegetation do not promote further lantana invasion into this habitat. Lantana is already widespread within UNP and HR. Preventing lantana from gaining a foothold in new areas, and reducing the number of plants already in place, is critical to maintaining and increasing the carrying capacity of the protected areas. The great abundance of P. maximum within the protected areas raises questions of whether removal of this invasive species is worth the expense, given the artificial nature of the grassland in general and its importance in the elephant diet.

39 Further review of the effects of P. maximum on the ecosystem need to be conducted to decide this issue.

Exclusion of all illegal grazing should be a top priority for protected-area managers. Livestock and elephants are using the same habitat and possibly competing for resources. Abundance of P. maximum was the strongest predictor of elephant presence in HR. If livestock prevent the recovery of this grass following a burn, it could reduce the amount of forage available for elephants. In UNP, a fence offers a clear and defining line between public and protected areas. The creation of fence along the south border of HR in addition to patrols by reserve officials would offer a chance for grasses to recover post burn and provide more forage for the elephants, especially during the wet season. Creation of waterholes within HR would also provide more resources for elephants during the dry season and would be especially beneficial if livestock were actively excluded from the reserve. These recommendations will prove useful not only for habitat management in Sri Lanka but also for other areas of the elephant range where similar ecosystems occur and where lantana and illegal grazing are concerns.

APPENDICES

Appendix 1

Figure 1.1: Location of study areas in Sri Lanka.

42 Appendix 2

Figure 2-1: The four main habitat types described during the dung and vegetation surveys within the study sites a) grassland, b) scrub savanna, c) scrub, and d) forest. All four habitat types were surveyed at Udawalawe National park but only grassland and scrub savanna were surveyed at Hurulu Forest Reserve.

43 Appendix 3

Total Length of Dung Number of Protected Area Habitat Transects (m) Vegetation Plots Burn Status UNP Grassland 5875 35 Unburnt UNP Scrub Savanna 8795 49 Unburnt UNP Scrub 3680 23 Unburnt UNP Forest 4550 26 Unburnt UNP Scrub Savanna 900 7 Burnt HR Grassland 1900 23 Unburnt HR Scrub Savanna 500 7 Unburnt HR Grassland 1500 18 Burnt HR Scrub Savanna 900 12 Burnt

Figure 3-1: The four main habitat types described during the dung and vegetation surveys within the study sites a) grassland, b) scrub savanna, c) scrub, and d) forest. All four habitat types were surveyed at Udawalawe National park but only grassland and scrub savanna were surveyed at Hurulu Forest Reserve.

44 Appendix 4

Figure 4-1: Vegetation plot design.

45 Appendix 5

Plant name Taxonomic status Accepted name Accepted author Accepted family Source Abutilon indicum Accepted Abutilon indicum (L.) Sweet Malvaceae tropicos;usda Azadirachta indica Accepted Azadirachta indica A. Juss. Meliaceae tropicos;usda Bauhinia racemosa Accepted Bauhinia racemosa Lam. Fabaceae tropicos Carissa spinarum Accepted Carissa spinarum L. Apocynaceae tropicos Cassia fistula Accepted Cassia fistula L. Fabaceae tropicos;usda Catunaregam spinosa Accepted Catunaregam spinosa (Thunb.) Tirveng. Rubiaceae tropicos Cordia dichotoma Accepted Cordia dichotoma G. Forst. Boraginaceae tropicos;usda Crotalaria laburnifolia Accepted Crotalaria laburnifolia L. Fabaceae tropicos;usda Croton bonplandianus Accepted Croton bonplandianus Baill. Euphorbiaceae tropicos;usda Croton officinalis Accepted Croton officinalis (Klotzsch) Alston Euphorbiaceae tropicos Diospyros ebenum Accepted Diospyros ebenum J. Koenig Ebenaceae tropicos Drypetes sepiaria Accepted Drypetes sepiaria (Wight & Arn.) Pax & K. Hoffm. Putranjivaceae tropicos Eucalyptus camaldulensis Accepted Eucalyptus camaldulensis Dehnh. Myrtaceae tropicos;usda Ficus benghalensis Accepted Ficus benghalensis L. Moraceae tropicos;usda Flacourtia inermis Accepted Flacourtia inermis Roxb. Salicaceae tropicos;usda Flueggea leucopyrus Accepted Flueggea leucopyrus Willd. Phyllanthaceae tropicos Gmelina asiatica Accepted Gmelina asiatica L. Lamiaceae tropicos;usda Hibiscus micranthus Accepted Hibiscus micranthus L. f. Malvaceae tropicos Imperata cylindrica Accepted Imperata cylindrica (L.) Raeusch. Poaceae tropicos Lannea coromandelica Accepted Lannea coromandelica (Houtt.) Merr. Anacardiaceae tropicos Lantana camara Accepted Lantana camara L. Verbenaceae tropicos;usda Lepisanthes sp. Accepted Lepisanthes Blume Sapindaceae tropicos Madhuca longifolia Accepted Madhuca longifolia (J. Koenig ex L.) J.F. Macbr. Sapotaceae tropicos Manilkara hexandra Accepted Manilkara hexandra (Roxb.) Dubard Sapotaceae tropicos Mimosa pudica Accepted Mimosa pudica L. Fabaceae tropicos;usda Mitragyna parvifolia Accepted Mitragyna parvifolia (Roxb.) Korth. Rubiaceae tropicos Morinda coreia Accepted Morinda coreia Buch.-Ham. Rubiaceae tropicos Murraya koenigii Accepted Murraya koenigii (L.) Spreng. Rutaceae tropicos;usda Pterospermum suberifolium Accepted Pterospermum suberifolium (L.) Willd. Malvaceae tropicos Sapindus emarginatus Accepted Sapindus emarginatus Vahl Sapindaceae tropicos Schleichera oleosa Accepted Schleichera oleosa (Lour.) Merr. Sapindaceae tropicos Sida sp. Accepted Sida L. Malvaceae tropicos Sida acuta Accepted Sida acuta Burm. f. Malvaceae tropicos;usda Sida cordifolia Accepted Sida cordifolia L. Malvaceae tropicos;usda Sida rhombifolia Accepted Sida rhombifolia L. Malvaceae tropicos;usda Strychnos potatorum Accepted Strychnos potatorum L. f. Loganiaceae tropicos Syzygium cumini Accepted Syzygium cumini (L.) Skeels Myrtaceae tropicos;usda Tectona grandis Accepted Tectona grandis L. f. Lamiaceae tropicos;usda Tephrosia purpurea Accepted Tephrosia purpurea (L.) Pers. Fabaceae tropicos;usda Urena sinuata Accepted Urena sinuata L. Malvaceae tropicos;usda Ziziphus oenopolia Accepted Ziziphus oenopolia (L.) Mill. Rhamnaceae tropicos Vitex altissima Accepted Vitex altissima L. f. Lamiaceae tropicos Allophylus zeylanicus Accepted Allophylus zeylanicus L. Sapindaceae WCSP Canthium coromandelicum Accepted Canthium coromandelicum (Burm.f.) Alston Rubiaceae WCSP Dimorphocalyx glabellus Accepted Dimorphocalyx glabellus Thwaites Euphorbiaceae WCSP Diospyros ovalifolia Accepted Diospyros ovalifolia Wight Ebenaceae WCSP Diplodiscus verrucosus Accepted Diplodiscus verrucosus Kosterm. Malvaceae WCSP Premna tomentosa Accepted Premna tomentosa Willd. Lamiaceae WCSP Eupatorium odoratum Synonym Chromolaena odorata (L.) R.M. King & H. Rob. Asteraceae tropicos Phyllanthus polyphyllus Synonym Diasperus polyphyllus (Willd.) Kuntze Euphorbiaceae tropicos Syzygium gardneri Synonym Eugenia gardneri (Thwaites) Bedd. Myrtaceae tropicos Grewia orientalis Synonym Grewia picta var. picta Baill. Malvaceae tropicos Adina cordifolia Synonym Haldina cordifolia (Roxb.) Ridsdale Rubiaceae tropicos Panicum maximum Synonym Megathyrsus maximus (Jacq.) B.K. Simon & S.W.L. Jacobs Poaceae tropicos Hyptis suaveolens Synonym Mesosphaerum suaveolens (L.) Kuntze Lamiaceae tropicos Vicoa indica Synonym Pentanema indicum var. indicum (L.) Ling Asteraceae tropicos Derris parviflora Synonym Pterocarpus parviflorus (Benth.) Kuntze Fabaceae tropicos Cassia siamea Synonym Senna siamea (Lam.) H.S. Irwin & Barneby Fabaceae tropicos Salvia reticulata Synonym Salvia glechomifolia M. Martens & Galeotti Lamiaceae WCSP

Figure 5-1: Taxonomic names of plant species recorded within the two study sites

46 Appendix 6

Protected Area Ground classification Satellite classification % Accurate Grassland (200163 m2) Forest (3780 m2) 73% UNP Grassland (275390 m2) Scrub (64012 m2) Other (7434 m2) Grassland (210874 m2) Forest (0 m2) * UNP Scrub Savanna (388167 m2) Scrub (141255 m2) Other (36038 m2) Grassland (13105 m2) Forest (3087 m2) 86% UNP Scrub (167338 m2) Scrub (144405 m2) Other (6741 m2) Grassland (11341 m2) Forest (154423 m2) 74% UNP Forest (208291 m2) Scrub (42087 m2) Other (441 m2) Grassland (113911 m2) Forest ( 3339 m2) 81% HR Grassland (139869 m2) Scrub (17830 m2) Other (4788 m2) Grassland (64642 m2) Forest (4977 m2) * HR Scrub Savanna (77747 m2) Scrub (5607m2) 2 Other (2520 m )

Figure 6-1: A comparison between the on–the-ground classification of habitats established at the start of the project and the habitat classification made by the

Smithsonian Institution using the software program eCognition© based on satellite imagery. Four habitats were defined from the ground surveys in Udawalawe National

47 Park (UNP): grassland, scrub savanna, scrub, and forest. Two habitats were defined from the ground surveys in Hurulu Forest Reserve (HR): grassland and scrub savanna. The software program only distinguished three habitat types from the satellite imagery: grassland, scrub, and forest. The “other” category incorporates bare ground, water, buildings, and floodplain landcovers.

48 Appendix 7

Elephant Livestock Mean distance % Grassland % Scrub % Forest % Water Dung Dung to water

Elephant Dung 1.00 0.18 0.01 0.19 -0.33 0.02 0.04 Livestock Dung 0.18 1.00 0.08 0.06 -0.23 0.02 -0.01 % Grassland 0.01 0.08 1.00 -0.69 -0.39 -0.05 0.06 % Scrub 0.19 0.06 -0.69 1.00 -0.31 -0.02 -0.05 % Forest -0.33 -0.23 -0.39 -0.31 1.00 -0.03 -0.11 % Water 0.02 0.02 -0.05 -0.02 -0.03 1.00 0.04 Mean distance to water 0.04 -0.01 0.06 -0.05 -0.11 0.04 1.00

Figure 7-1: Correlation matrix of landscape scale variables in Udawalawe National Park.

Elephant Livestock Mean distance % Medium % Grassland % Scrub % Forest % C. imperata % Lantana % P. maximum % Short Grass % Teak Dung Dung to water Grass Elephant Dung dung 1.00 0.23 0.12 -0.01 -0.29 0.04 -0.11 0.04 0.05 -0.07 0.10 -0.05 Livestock Dung 0.23 1.00 0.05 0.05 -0.24 0.10 0.03 -0.23 -0.11 0.20 0.03 0.07 % Grassland 0.12 0.05 1.00 -0.85 -0.30 -0.01 -0.07 -0.21 -0.08 0.39 -0.23 0.30 % Scrub -0.01 0.05 -0.85 1.00 -0.18 -0.09 0.11 0.24 0.20 -0.20 0.15 -0.32 % Forest -0.29 -0.24 -0.30 -0.18 1.00 -0.10 -0.06 -0.04 -0.15 -0.28 0.17 -0.12 Mean distance to water 0.04 0.10 -0.01 -0.09 -0.10 1.00 -0.03 0.02 -0.10 -0.23 0.20 0.00 % C. imperata -0.11 0.03 -0.07 0.11 -0.06 -0.03 1.00 -0.05 0.13 -0.13 -0.07 -0.10 % Lantana 0.04 -0.23 -0.21 0.24 -0.04 0.02 -0.05 1.00 -0.11 -0.18 0.06 -0.18 % Medium Grass 0.05 -0.11 -0.08 0.20 -0.15 -0.10 0.13 -0.11 1.00 -0.24 -0.09 -0.18 % P. maximum -0.07 0.20 0.39 -0.20 -0.28 -0.23 -0.13 -0.18 -0.24 1.00 -0.39 0.35 % Short Grass 0.10 0.03 -0.23 0.15 0.17 0.20 -0.07 0.06 -0.09 -0.39 1.00 -0.27 % Teak -0.05 0.07 0.30 -0.32 -0.12 0.00 -0.10 -0.18 -0.18 0.35 -0.27 1.00

Figure 7-2: Correlation matrix of landscape- and local-scale variables in Udawalawe National Park. Local-scale variables include plant and species data derived from the vegetation plots.

Elephant Livestock % Bare % Grassland % Scrub % Burn Area Dung Dung Ground Elephant Dung 1.00 -0.02 0.07 -0.08 0.02 0.01 Livestock Dung -0.02 1.00 -0.19 0.27 -0.05 0.34 % Grassland 0.07 -0.19 1.00 -0.79 -0.10 -0.06 % Scrub -0.08 0.27 -0.79 1.00 0.00 0.10 % Bare Ground 0.02 -0.05 -0.10 0.00 1.00 -0.14 % Burn Area 0.01 0.34 -0.06 0.10 -0.14 1.00

Figure 7-3. Correlation matrix of landscape scale variables in Hurulu Forest Reserve.

Elephant Livestock % Bare % Burn % C. % Short % Medium % Grassland % Scrub % Lantana % Bare % P. maximum Dung Dung Ground Area imperata Grass Grass Elephant Dung 1.00 0.01 0.13 -0.11 -0.01 0.06 -0.10 -0.01 -0.19 -0.03 -0.03 0.31 Livestock Dung 0.01 1.00 -0.16 0.25 -0.05 0.34 -0.08 0.13 -0.33 -0.06 0.02 0.04 % Grassland 0.13 -0.16 1.00 -0.81 -0.13 -0.04 0.04 -0.11 0.00 0.12 0.05 0.01 % Scrub -0.11 0.25 -0.81 1.00 0.06 0.08 -0.02 0.11 -0.13 -0.10 -0.05 0.02 % Bare Ground -0.01 -0.05 -0.13 0.06 1.00 -0.13 -0.05 -0.17 0.18 -0.03 -0.09 -0.08 % Burn Area 0.06 0.34 -0.04 0.08 -0.13 1.00 -0.20 0.68 -0.49 0.17 -0.11 -0.01 % Lantana -0.10 -0.08 0.04 -0.02 -0.05 -0.20 1.00 -0.13 -0.07 -0.06 0.03 0.08 % Bare -0.01 0.13 -0.11 0.11 -0.17 0.68 -0.13 1.00 -0.22 0.26 -0.41 -0.07 % P. maximum -0.19 -0.33 0.00 -0.13 0.18 -0.49 -0.07 -0.22 1.00 -0.23 -0.34 -0.27 % C. imperata -0.03 -0.06 0.12 -0.10 -0.03 0.17 -0.06 0.26 -0.23 1.00 -0.12 -0.06 % Short Grass -0.03 0.02 0.05 -0.05 -0.09 -0.11 0.03 -0.41 -0.34 -0.12 1.00 0.19 % Medium Grass 0.31 0.04 0.01 0.02 -0.08 -0.01 0.08 -0.07 -0.27 -0.06 0.19 1.00

Figure 7-4: Correlation matrix of landscape and local scale variables in Hurulu Forest Resreve.

Appendix 8

Grassland Scrub Savanna Scrub Forest

Figure 8-1: The first two axes of the principle components analysis of the grass species,

T. grandis, and L. camara in Udawalawe National Park.

Appendix 9

Figure 9-1: The first two axes from the principle components analysis of the grass species, L. camara and bare ground, the major environmental characteristics in Hurulu

Forest Reserve.

54 References

BARNES, R. 2001. How reliable are dung counts for estimating elephant numbers? African Journal of Ecology 39:1-9.

BARNES, R. F. W. & JENSEN, K. L. 1987. How to count elephants in forests. IUCN African elephant & Rhino specialist group Technical Bulletin 1:1–6.

BARNES, R. F. W., ASAMOAH-BOATENG, B., NAADA MAJAM, J. & AGYEI-OHEMENG, J. 1997. Rainfall and the population dynamics of elephant dung-piles in the forests of southern Ghana. African Journal of Ecology. 35: 39-52 BLAKE, S. & HEDGES, S. 2004. Sinking the flagship: the case of forest elephants in Asia and Africa. Conservation Biology 18:1191–1202.

BLANC, J.J., BARNES, R.F.W., CRAIG, G.C., DUBLIN, H.T., THOULESS, C.R., DOUGLAS- HAMILTON, I., & HART, J.A. 2007. African elephant status report 2007.

BUCKLAND, S. T., ANDERSON, D. R., BURNHAM, K. P., LAAKE, J. L., BORCHERS, D. L. & THOMAS, L. 2004. Advanced distance sampling: estimating abundance of biological populations. Oxford University Press, Oxford. 434 pp.

CENTRAL BANK OF SRI LANKA. 2012. Annual report 2012. Colombo, Sri Lanka.

CHOUDHURY, A., LAHIRI CHOUDHURY, D.K., DESAI, A., DUCKWORTH, J.W., EASA, P.S., JOHNSINGH, A.J.T., FERNANDO, P., HEDGES, S., GUNAWARDENA, M., KURT, F., KARANTH, U., LISTER, A., MENON, V., RIDDLE, H., RUBEL, A. & WIKRAMANYAKE, E. 2008. Elephas maximus. IUCN red list of threatened species. Version 2011.1

DAY, M.D., WILEY, C.J., PLAYFORD, J. & ZALUCKI, M.P. 2003. Lantana camara: Current Management Status and Future Prospects. ACIAR Monograph Series, Canberra. de SILVA, S., RANJEEWA, A.D.G. & WEERAKOON, D. 2011. Demography of Asian elephants (Elephas maximus) at Uda Walawe National Park, Sri Lanka based on identified individuals. Biological Conservation 144:1742–1752

DUGGIN, J.A. & GENTLE, C.B. 1998. Experimental evidence on the importance of disturbance intensity for invasion of Lantana camara L. in dry rainforest-open ecotones in north-eastern NSW, Australia. Forest Ecology and Management 109:297-30.

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

55 FERNANDO, P. 1997. Keeping jumbo afloat: is translocation an answer to the human- elephant conflict? Sri Lanka Nature 1:4-12

FERNANDO, P. 2000. Elephants in Sri Lanka: past, present, and future. Loris 22:38-44.

FERNANDO, P., WIKRAMANAYAKE, E., WEERAKOON, D., JAYASINGHE, L.K.A., GUNAWARDENA, M. & JANAKA, H.K. 2005. Perceptions and patterns of human-elephant conflict in old and new settlements in Sri Lanka: insights for mitigation and management. Biodiversity and Conservation 14:2465-2481.

FERNANDO, P., WIKRAMANAYAKE, E., WEERAKOON, D., JANAKA, H.K., GUNAWARDENA, M., JAYASINGHE, L.K.A., NISHANTHA, H.G. & PASTORINI, J. 2006. The future of Asian elephant conservation: Setting sight beyond protected area boundaries. Pp. 252-260 in McNeely, J.A., McCarthy, T.M., Smith, A., Olsvig-Whittaker, L. & WIKRAMANAYAKE, E.D. (Eds.) Conservation Biology in Asia. Society for Conservation Biology Asia Section and Resources Himalaya Foundation, Kathmandu, Nepal.

FERNANDO, P. & LEIMGRUBER, P. 2011. Are Asian elephants a keystone or edge species in dry forest. Pp. 151-163 in McShea, W.J, Davies, S.J., Phumpakphan, N. & Pattanavibool, A. (Eds.) The Ecology and Conservation of Seasonally Dry Forests in Asia. Smithsonian Institution Scholarly Press.

FLINT, E.P. 1994. Changes in land-use in south and southeast-Asia from 1880 to 1980—A data-base prepared as part of a coordinated research program on carbon fluxes in the tropics. Chemosphere 29:1015-1062.

GADGIL, M. & MEHER-HOMJI, V.M. 1985. Land use and productive potential of Indian savannas. Pp. 107-113 J.C. Tothill and J.J. Mott (Eds.) Ecology and Management of the World’s Savannas. Australian Academy of Science. Canaberra.

GENTLE, C.B. & DUGGIN, J.A. 1997. Lantana camara L. invasion in dry rainforest ecotones: the role of disturbance associated with fire and cattle grazing. Australian Journal of Ecology 22:298-306.

GOODEN, B., FRENCH, K. O. & TURNER, P. 2009. Invasion and management of a woody plant, Lantana camara L., alters vegetation diversity within wet sclerophyll forest in southeastern Australia. Forest Ecology and Management 257:960-967.

GUY, P.R.1975. The daily food intake of the African elephant, Loxodonta africana Blumenbach, in Rhodesia. Arnoldia 7:1-8.

HEDGES, S. 2006. Conservation. Pp. 475-490 in Fowler, M. E. & Mikota, S. K. (Eds.) Biology, Medicine and Surgery of Elephant,. Blackwell Publishing, Oxford, UK.

HIREMATH, A.J. & SUNDARAM, B. 2005. The fire-Lantana camara cycle hypothesis in Indian forests. Conservation and Society 3:26-42.

JULIEN, M.H. & GRIFFITHS, M.W. 1998. Biological Control of Weeds. A Worldwide Catalogue of Agents and their Target Weeds, Fourth ed. CAB Publishing, CAB International, Wallingford, UK.

KOH, L.P. & WILCOVE, D.S. 2008. Is oil palm agriculture really destroying tropical biodiversity? Conservation Letters 1:60–64.

LEIMGRUBER, P., GAGNON, J.B., WEMMER, C., KELLY, D.S., SONGER, M.A. & SELIG, E.R. 2003. Fragmentation of Asia's remaining wildlands: implications for Asian elephant conservation. Animal Conservation 6:347-359.

LOWE S., BROWNE M., BOUDJELAS S., De POORTER M. 2004. 100 of the World’s Worst Invasive Alien Species: A selection from the Global Invasive Species Database. The Invasive Species Specialist Group (ISSG) a specialist group of the Species Survival Commission (SSC) of the World Conservation Union (IUCN), 12pp.

McKAY, G.M., 1973. The ecology and behaviour of the Ceylon elephant in southeastern Ceylon. In: Eisenberg, J.F., M.G.M., Seidensticker, J. (Ed.), Asian Elephants. Friends of the National Zoo and National Zoological Park (Smithsonian Institute), Washington, DC.

MISHRA H. R. 1982. Balancing Human Needs and Conservation in Nepal's Royal Chitwan Park. Ambio 11: 246-251.

MOLEGODA, S. 1984. Ecology of Uda Walawe National Park. MSc thesis, University of Colombo

NCHANJI, A.C. & PLUMPTRE, A. 2001. Seasonality in elephant dung decay and implications for censusing and population monitoring in southwestern Cameroon. African Journal of Ecology 39:24-32.

NORRIS, C.E. 1959. Preliminary report on the Ceylon elephant field survey. Wild Life Protection Society, Colombo.

ODADI, W.O. 2011. African wild ungulates compete with or facilitate cattle depending on season. Science 334:594-594 PASTORINI, J., NISHANTHA, H. G., & FERNANDO, P. 2007. A Preliminary Study of Dung Decay in the Yala National Park, Sri Lanka. Gajah 27:48-51.

PERERA, B.M.A.O. 2009. The human–elephant conflict: a review of current status and mitigation methods. Gajah 30:41–52.

R CORE Team. 2013. R: A language and environment for statistical computing. R Foundation for Statisitcal Computing, Vienna, Austria. URL http://www.R-project.org/.

RANWALA, S.M.W. & THUSHARI, P.G.I. 2012.Current status and management options for invasive plants at the Mihintale Wildlife Sanctuary. Journal of the National science Foundation of Sri Lanka 40:67-76

SANTIAPILLAI, C. & JACKSON, P. 1990. The Asian elephant: an action plan for its conservation. World Conservation Union/Species Survival Commission Asian Elephant Specialist Group, Gland, Switzerland.

SCHMERBECK, J. & SEELAND, K. 2007. Fire supported forest utilisation of a degraded dry forest as a means of sustainable local forest management in Tamil Nadu South India. Land Use Policy 24:62-71.

SCHULZ, T. T. & LEININGER W. C. 1990. Differences in riparian vegetation structure between grazed areas and exclosures. Journal of Range Management 43:295-299.

SHARMA, G.P., RAGHUBANSHI, A.S., & SINGH, J.S. 2005. Lantana camara invasion: an overview. Weed Biology and Management 5:157-167.

SHOSHANI, J. & EISENBERG, J. F.. 1982. Elephas maximus. Mammalian Species 182:1–8.

SODHI, N.S., KOH, L.P., BROOK, B.W. & NG, P.K.L. 2004. Southeast Asian biodiversity: an impending disaster. Trends in Ecology and Evolution 19: 654-660.

SUKUMAR, R. 1989. The Asian elephant: ecology and management. Cambridge University Press, Cambridge.

SUKUMAR, R. 1990. Ecology of the Asian elephant in southern India. II. Feeding habits and crop raiding patterns. Journal of Tropical Ecology 6:33-53

58 SUKUMAR, R. 1992. The Asian elephant: ecology and management, 2nd edition. Cambridge University Press, Cambridge.

SUKUMAR, R. 2003. The Living Elephants. Oxford University Press, Oxford.

SZARO, R. C. 1989. Riparian forest and scrubland community types of Arizona and New Mexico. Desert Plants 9:69-138.

THOMAS, L., BUCKLAND, S.T., REXSTAD, E.A., LAAKE, J.L., STRINDBERG, S., HEDLEY, S.L., BISHOP, J. R.B., MARQUES, T. A., & BURNHAM, K. P. 2010. Distance software: design and analysis of distance sampling surveys for estimating population size. Journal of Applied Ecology 47:5-14.

URYU, Y., MOTT, C., FOEAD, N., YULIANTO, K., BUDIMAN, A., SETIABUDI, TAKAKAI, F., NURSAMSU, S., PURASTUTI, E., FADHLI, N. , HUTAJULU, C.M.B., JAENICKE, J., HATANO, R., SIEGERT, F. & STUWE, M. 2008. Deforestation, forest degradation, biodiversity loss, and CO2 emissions in Riau, Sumatra, Indonesia. WWF Indonesia Technical Report. http://assets.panda.org/downloads/ riau_co2_report__wwf_id_27feb08_en_lr_.pdf; accessed 11/9/2013. van OOSTERHOUT, E. 2004. Lantana control manual: Current management and control of lantana (Lantana camara) in Australia, Queensland Department of Natural Resources, Mines and Energy, Brisbane. van WILGREN, B., GOVENDER, N., BIGGGS, H., NTSALA, D. & FUNDA, X. 2004. Response of savanna fire regimes to changing fire-management policies in a large African national park. Conservation Biology 18:1533–1540.

VITOUSEK, P. M. , LOOPE, L.L., & STONE, C.P. 1987. Introduced species in Hawaii: Biological effects and opportunities for ecological research. Trends in Ecology & Evolution 2:224-227.

VIVIAN-SMITH, G., & PANETTA, F.D. 2009. Lantana (Lantana camara) seed banks dynamics: seedling emergence and seed survival. Invasive Plant Science and Management 2:141-150.

WATERHOUSE, D.F. & NORRIS, K.R. 1987. Biological Control: Paciﬁc Prospects. Inkata Press, Melbourne

WHITE, L.J.T. 1995. Factors affecting the duration of elephant dung piles in rain forest in the Lopé Reserve, Gabon. African Journal of Ecology 33: 142-150.

WILSON, G., Desai, A.A., Dalice A. Sim and Wayne L. Linklater. 2013. The influence of the invasive weed Lantana camara on elephant habitat use in Mudumalai Tiger Reserve, southern India. Journal of Tropical Ecology 29:199-207.

WEERAKOON, D.K, GUNAWARDENE, M.D., JANAKA, H.K, JAYASINGHE, L.K.A., PERERA, R.A.R., FERNANDO, P. & WIKRAMANAYAKE, E. 2004. Ranging behavior and habitat use of elephants in Sri Lanka. Pp. 68-70 in Jayewardene, J (Ed.) Endangered Elephants: Past Present and Future. Sri Lanka, Biodiversity & Elephant Conservation Trust.

WIKRAMANAYAKE, E.D., HATHURUSINGHE, H.S., JANAKA,H.K., JAYASINGHE, L.K.A., FERNANDO, P., WEERAKOON, D.K., GUNAWARDENE, M.D. 2004. The human-elephant conflict in Sri Lanka: lessons from mitigation, management, and conservation from traditional land-use patterns. Pp 164-169 in Jayewardene, J (Ed.) Endangered Elephants: Past Present and Future. Sri Lanka, Biodiversity & Elephant Conservation Trust.

ZALUCKI, M. P., DAY, M. D., & PLAYFORD, J. 2007. Will biological control of Lantana camara ever succeed? Patterns, processes & prospects. Biological Control 42, 251–261.

ZUBAIR, L., SIRIWARDHANA, M., CHANDIMALAB, J., & YAHIYAB, Z. 2008. Predictability of Sri Lankan rainfall based on ENSO. International Journal of Climatology 28: 91–101.

60 CHAPTER THREE

LANTANA camara: ALTERING FIRE IN AN ECOSYSTEM

Lantana camara, an invasive species, can severely alter a landscape in terms of structure, composition and function, creating a marked change in the fire regime of the invaded habitat. As one of the world’s top 100 invaders, this plant is poised to wreak havoc on sensitive and critical habitats throughout many semi-tropical and tropical habitats. This paper explores current research and ideas surrounding the proliferation of

L. camara, especially in relation to fire, and posits several areas where more work would be helpful in the quest to understand and prevent L. camara invasion.

Exotic plants around the world

Invasive species present one of the most dire challenges faced by conservationists today. The numerous threats exotic invasive species pose to a new environment are well documented. Changes in structure [1], composition [2] and function [3] of a habitat can have resounding environmental and economic impacts [4]. These changes can also have an enormous effect on a habitat’s susceptibility to fire and alter the landscape’s fire regime over time.

Exotic grasses are often cited as the poster species for the invasion of fire- favored plants into native ecosystems. The presence of grasses in an environment can increase the fire frequency, rate of spread, and the extent of a burn [5,6]. Grasses are well adapted to survival in fire prone areas and their presence can lead to shorter

61 intervals between fires, encouraging the proliferation of the introduced grass species to the detriment of native grasses and woody perennials. This grass-fire cycle has been well documented throughout the western United States, and the Hawaiian Islands [7].

Tree and shrub species, however, can be equally adept at creating a similar self- propagating fire cycle. Studies in the fynbos of South Africa have documented that invasive Hakea sericea can increase fuel loads and decrease average fuel moisture [8]. In total, Van Wilgren et al. [8] saw a 3 to 10 fold increase in above ground-biomass from all woody plant invaders. Unsurprisingly, these invasions also lead to greater fire intensities in invaded areas versus pristine environments [8]. These exotic shrubs can alter the hydrology of the area in favor of fire as well. Invaded areas of the fynbos had up to a

50% reduction in stream flow due to increased evaporation by the increased above- ground biomass [8]. This decline in rainfall passing through the midstory to reach the ground can decrease soil moisture, resulting in greater fire severity when a burn does occur.

Lantana camara

Origin

L. camara, is native to Central and South America, and it has spread throughout many regions in Asia, Australia and Africa after being introduced as an ornamental plant [4].

Early accounts of its invasiveness claim it spread from a single location in India contained within a few square kilometers out as much as 40 kilometers in all directions,

62 covering farms, grassland and forest [9]. L. camara can grow as an individual plant or into a thick hedge, crowding out native species, especially in areas of disturbance. L. camara was one of the first plants ever targeted for biological control, but little progress has been made at containing the species [10]. Efforts for biological control began in

1902 but have not been successful despite the release of 41 agents in 40 different regions [11,12]. L. camara produces edible berries year-round, which are eaten by for many frugivores [9] allowing for continuous dispersal of seeds. In addition, the foliage of the plant contains pentacyclic triterpenoids, a toxin that has led to many livestock deaths [13].

Lantana camara and fire

It has been hypothesized that L. camara creates a positive fire feedback loop [14,15,16] or a L. camara-fire cycle similar to the grass-fire cycle (Figure 1). L. camara readily establishes in highly disturbed areas, from areas such as agriculture and plantations

[17] to roadsides [18]. Once it has invaded, it stimulates higher fire frequency and intensity, causing more disturbances within the ecosystem, thus repeating the cycle

[16]. Early accounts of L. camara infestations allude to the extreme intensity of a L. camara fire [19]. As the cycle continues, L. camara is able to expand its range by colonizing the edges of the fire and as seed distributors deposit its seeds in novel and newly disturbed habitats.

63 Figure 1. A flow chart describing the Lantana camara-fire cycle. From Hiremath and Sundaram [16].

L. camara promotes flammability within an ecosystem primarily by dramatically altering the structure of the vegetation. Not only does it increase the biomass within its invaded range, but it creates a continuous fuel bed in the lower to mid stories of the environment horizontally as well as vertically [15] (Figure 2). Unlike many woody species, L. camara burns easily even when green [20]. Increasing flammable biomass on the ground level of an ecosystem correlates positively with fire frequency and extent. L. camara can also produce climbing stems of up to 35 feet [21,22], bridging the gap between the mid-story and crown in an invaded forest. This fuel ladder, in addition to the extreme flammability of L. camara, can produce devastating crown fires, reducing a mature forest to a sea of snags [19, 23]. A study on Forty Mile Scrub National

64 A

Figure 2. Changes to forest environments following a Lantana camara invasion. Community A has no midstory, reducing the chance of a crown fire, and a discontinuous fuel bed. Community B displays fuel ladders to the canopy and a more continuous fuel bed throughout the surface and midstory levels, with an increase of above ground biomass.

65 Park, Australia determined that the most destructive fires to occur in the park all came after the time since the park had been invaded by L. camara [14].

Berry et al. [23] found that forests invaded by L. camara in Austraila had greater proportion of medium sized fuels compared to the other size classes. They postulated that due to the higher heartwood to pith ratio, these fuels would increase fire intensity and residence time. This increase would result in a greater percentage of native plant deaths, especially of dry rainforest trees [14, 23]. And while Berry et al. [23] determined individual stems of L. camara may not be more flammable when comparing between it and those of native vegetation, the sheer mass of this medium sized fuel in the understory can be enough to encourage more frequent fires.

Post fire regeneration of Lantana camara

Seed regeneration

Chemical signals produced by burning vegetation can initiate and enhance seed germination in many plant species [24, 25]. Due to their size, L. camara seeds tend to fall deeper into the litter layer, escaping direct flame exposure but coming into contact with smoke [25]. In their study assessing the use of fire as a control for invasive populations of L. camara, Raizada and Rahubanshi [25] found that smoke exposure leads to a faster germination shortening germination start times from a couple of months to a couple weeks [25]. They also saw a significant increase in successful germination, seedling recruitment of L. camara, and overall healthier plants. Raizada

66 and Rahubanshi [25] postulated that the chemicals found in smoke may increase the permeability of L. camara seeds, increasing water absorption and initiate germination.

Based on their results, they recommended against the use of fire as a control method.

Seed germination is also stimulated by increased light intensity and higher soil temperatures [26]. Fire can be a great advantage for L. camara especially if the native vegetation does not have soil-based seed banks. After sprouting, L. camara is able to quickly reach the stage where it can produce viable seeds, which it does in large quantities. Seeds that aren’t consumed by herbivores are stored in soil seed banks. Fire both increases soil temperatures and removes above ground biomass shading the seeds banked in the soil. Seed banks enable L. camara to exploit areas after a disturbance, especially in larger burn sites which may have fewer standing trees for birds to perch on and drop competitor seeds [26].

Vegetative regeneration

In addition to seed-related fire survival strategies, L. camara has other several traits that allow it to exploit the environment and survive after a disturbance such as a fire. Fire removes light competitors for seedlings and increases nutrient availability through ash deposits [26]. L. camara has a relatively shallow root system compared to the woody plant species it competes with for nutrients [26]. This allows it to access the nutrients faster than seedling of competing species, establishing its population in recently burnt habitat sooner. L. camara has a very efficient nutrient uptake system and can therefore establish itself on even severely degraded soil.

67 Mature L. camara plants have several adaptations which circumvent many proposed control methods. Any horizontal stem of a plant can produces a root when it comes in contact with the soil, effectively creating a clone of the original plant [27].

Similarly, if a stem becomes covered by soil or a buildup of decaying vegetation, it too can root and form another plant. Even uprooted plants removed by contractors or land- owners can resprout at their dumping grounds if not properly disposed of [12]. Several researchers have noted that L. camara regenerates denser than the orginal plant was prior to the disturbance [1,28].

Any disturbances that remove competing vegetation allows established L. camara plants to grow faster under increased light availability [1]. Duggin and Gentle also saw that if a nutrient pulse was given in addition to an increased availability of light, an even greater increase in growth was observed. This was even seen when the ground level vegetation was left undisturbed and only the canopy was opened [1].

Allelopathy

L. camara builds upon its advantage gained early establishment by using allelopathy against competing plants [29,30]. Allelopathy is a form of interference competition defined by Inderjit and Calloway as the negative effect one competitor has on another through the release of chemical compounds into the environment[31]. Allelochemicals may be found throughout a plant including roots, rhizomes, leaves, stems, pollen, seeds and flowers, and are released by root exudation, leaching, volatilization, and/or by decomposition of plant material [32]. These effects are mediated by the amount,

68 strength, and rate of dispersal of the chemicals released; how the chemicals interact with the abiotic and biotic factors present in the environment; and how the neighboring

(target) species react to the chemicals [33].

Allelochemicals can affect the native community in three general ways: through direct inhibition, reducing available nutrients, or by reducing mycorrhizae in the area

[34]. In direct inhibition, allelopathic plants can outcompete their neighbors by releasing a compound toxic to these species, so that even if the resource acquisition abilities of the native plant exceeds that of the exotic, the allelochemicals prevent the natives from absorbing the resource [35]. L. camara [26], use chemicals such as phytotoxic (-) – catechin and 8-hydroxyquinoline to interfere with the resource retention, resource absorption, or other growth properties of surrounding plants [34]. Allelopathic plants can change the soil microbiota, and affect abiotic conditions, reducing available nutrients beyond the thresholds of where native plants can exist [35].

Lantana camara and development

L. camara is a light-dependent species, unable to survive in the low light conditions of a mature forest with a dense, intact canopy [28]. Given the extreme difficulty in removing an infestation once established and the devastating impacts this plant can have on the health of an ecosystem, any managers of an area with the potential for L. camara invasion should attempt to prevent fragmentation within the ecosystem as much as possible. Unfortunately, the primary zone of infestation is throughout the developing

69 world. This area is experiencing rapid growth, often with little thought to future management of the disturbed natural areas this growth is creating. For example, logging practices started in the developed nation such as thinning would only open the forest canopies, allowing more solar penetration and opening the habitat to invasive species. Agriculture is fragmenting once-continuous forest, creating edge habitat and exposing once pristine forest to invasion. In the case of L. camara, habitat loss and fragmentation and the associated are is in fact opening natural areas to invasion.

Herbivores and Lantana camara

Third world countries face many challenges not often encountered by biologists and conservationists from developed nations. When crafting habitat management plans for developing nations, scientists from developed countries often lack of basic knowledge of the local area where they are trying to model or cultivate conservation practices. One such local issue I believe may be having a huge impact on the proliferation of L. camara is the presence of cattle. Unlike in many North American or European locales, in developing countries such as tropical Asia cattle tend have free range throughout the landscape.

Gentle and Duggin [26] partially examined the role of cattle in the promoting the growth of L. camara in a dry rainforest ecotone in NSW, Australia. They determined that the biomass reduction and soil disturbance caused by cattle on an environment can increase L. camara success. This relationship was primarily driven by grazing which

70 reduced the above ground biomass, increasing light penetration to the soil and any L. camara seeds or seedlings.

From my observations in Sri Lanka, I suggest that a different aspect of the cattle’s usage of the environment may have an even greater effect on the spread of L. camara. Throughout many areas in India and Sri Lanka, ranchers illegally graze their cattle on protected areas. This raises two issues: 1) cattle effects on the spread of fire and 2) increasing human presence in the protected areas. Cattle trails are readily apparent in many of the national parks in Sri Lanka, yet no one has studied how these trails affect the spread of fire within the ecosystem. These ‘natural’ fire breaks can have a great impact if the cattle avoid L. camara thickets, but create a break in the fuel beds within uninfested areas.

The second issue cattle introduce into an ecosystem where they are grazing is the increased presence of humans as they are tending their herds. Most of the fires in

Sri Lanka are started by humans, either purposely as a method to scare game out of the forest or accidentally from an escaped cooking fire or cigarette butt. These events inflate the natural fire regime of the areas. Native vegetation may not be able to compete successfully for space and resources at this increased rate of disturbance, allowing invasive species such as L. camara in come into the environment.

Other herbivores can also contribute to the spread of L. camara. Despite the toxic chemicals contained by the L. camara foliage, many animals, including humans, consume its berries, possibly carrying the seeds hundreds of miles from the source plant

71 before defecating. In addition to transporting the seeds, animals many play a role in opening hospitable locations for seed germination. L. camara is prolific in many national parks throughout Asia, which are home to mega-herbivores, often cited as landscape engineers for the huge impacts they have on their environment. Even small disturbances such as tree falls allow L. camara to establish a foothold within a forest [36]. Asian elephants can easily and often uproot a tree creating a hole in the canopy, allowing for an invasion. Fensham et al. [14] similarly determined the presence of wild pigs in

Australian national parks can severely damage canopy trees and promote the spread of

L. camara.

Management Options

As prescribed fires of mild to moderate intensity appear to increase the chance of a successful L. camara invasion, active fire suppression in mature forests has been a suggested method of control for L. camara [26]. As it is not a matter of if a forest will burn, but when, this is not a realistic management plan. By excluding all fire, the fuel load would become high and lead to severe fires, destroying most native vegetation and allowing L. camara a clear path for invasion. And rarely are disturbances limited to only one factor benefiting L. camara. Gentle and Duggin [26] found that when two or more disturbances act together, the relative success of L. camara plants increases. While fire has the ability to provide multiple disturbances by itself (i.e., reduction of above ground biomass of competitor species and increased nutrients in the soil), its effects can be

72 amplified by many human and animal activities. Multiple disturbances, such as using fire to clear an area for agriculture and then fertilizing the soil, make for prime invasion sites. In addition, L. camara’s ability to produce seeds all year long means that fire will not significantly reduce a population’s ability to reproduce: without a season for fruit bearing, L. camara ensures that at least part of its population is always producing fruit.

[26].

Summary

L. camara is invasive in the tropical and semi-tropical regions of the world, areas that are now experiencing high rates population expansion and development of natural habitats into agriculture and industry. These activities will only continue to open up more land to invasion by this aggressive weed. Very little progress has been made in determining control methods for this highly successful invader once it has become established. Invasive species management needs to be at the forefront of every manager’s mind when they are establishing management plans for the natural resources of their country. Fire is a commonly used tool for habitat management throughout the world. L. camara has many traits which help it to survive fire and exploit the post fire environment. In areas where L. camara is present, applying fire to a landscape could be the start of a vicious cycle leading to the destruction of resource trying to be protected.

73 Literature Cited

1 Duggin, J.A. and Gentle, C.B (1998) Experimental evidence on the importance of disturbance intensity for invasion of Lantana camara camara in dry rainforest- open ecotones in north-eastern NSW, Australia. For. Eco. Manag.109, 297-30 2 Gooden, B. et al. (2009) Invasion and management of a woody plant, Lantana camara camara L.. alters vegetation diversity within wet sclerophyll forest in southeastern Australia. For. Eco. Manag. 257, 960-967 3 Vitousek, P. M. et al. (1987) Introduced species in Hawaii: Biological effects and opportunities for ecological research. Trends Ecol. Evol. 2, 224-227. 4 Mack, R.N. et al. (2000) Biotic invasions: causes, epidemolgy, global consequences and control. Ecol. Appl. 10, 689-710 5 Whisenant, S.G. (1990) Changing fire frequencies on Idaho’s Snake River Plain: ecological and management implications. U.S.D.A.. Forest Service, lntermountain Research Station 6 Smith, C.W. and Tunison, T. (1992) Fire and alien plants in Hawai’i: research and management implications for native ecosystems, in Alien Plant Invasions in Native Ecosyslems of Hawai’i: Management and Research (Stone, C.P. et al., eds), 394-408, University of Hawaii Press 7 Mack, M.C. and D’Antonio C.M. (1998) Impacts of biological invasions on disturbance regimes. Trends Ecol. Evol. 13, 195-198 8 Van Wilgen, B.W. and Richardson. D.M. (1985) The effects of alien shrub invaeions on vegetation structure. and fire behaviour in South African fynbos shrublands: a simulation study. J. Appl. Ecol. 22, 955-966 9 Hakimuddin, M. (1929) Lantana camara in northern India as a pest and its probable utility in solving the cowdung problem. Indian For. 56, 405-410. 10 Zalucki, M.P. et al. (2007) Will biological control of Lantana camara camara ever succeed? Patterns, processes & prospects. Bio. Con. 42, 251–261. 11 Julien, M.H. and M.W. Griffiths (eds.). (1998) Biological Control of Weeds. A Worldwide Catalogue of Agents and their Target Weeds, Fourth ed. CAB Publishing, CAB International, Wallingford, UK. 12 Day, M.D. et al. (2003) Lantana camara: Current Management Status and Future Prospects. ACIAR Monograph Series, Canberra. 13 Sharma, O.P. et al. (1981) Areview of the toxicity of Lantana camara camara (Linn) in animals. Clin. Tox. 18,1077-1094. 14 Fensham, R.J. et al. (1994) The invasion of Lantana camara camara in Forty Mile Scrub National Park, North Queenslands. Austr. J. Ecol. 19, 297-305 15 Brooks, M. et al. (2004) Effects of Invasive Alien Plants on Fire Regimes. BioSci.54, 677-688 16 Hiremath, A.J. and B. Sundaram. 2005. The fire-Lantana camara cycle hypothesis in Indian forests. Con. Soc. 3, 26-42.

74 17 Sharma, G.P. et al. (2005) Lantana camara invasion: an overview. Weed Bio. Manag. 5, 157-167 18 Munir K. (1996) Lantana: a weed. J. Sci. Indust. Res. 69, 524-527 19 Tireman H. (1916) Lanatana in the forest of Coorg. Indian For. 42, 384-392 20 Gujral G.S. and Vasudevan P. (1983) Lantana camara L. a problem weed. J. Sci. Indust. Res. 42, 281-286 21 Anon. (1895) Is Lantana a friend or an enemy? Indian For. 21, 454-460 22 Hyland H.P.(2003) Australian Tropical Rain Forest Plants: Trees, Shrubs and Vines. CSIRO, Melbourne. 2002 23 Berry et al. 2010 The invasive weed Lantana camara increases fire risk in dry rainforest by altering fuel beds. Weed Res. 51, 525–533 24 Baldwin et al. 1994 25 Raizada P. and Raghubanshi A.S. (2010) Seed germination behavior of Lantana camara in response to smoke J. Trop. Eco. 51, 347-352 26 Gentle C.B. and Duggin J.A. (1997) Lantana camara L. invasion in dry rainforest ecotones: the role of disturbance associated with fire and cattle grazing. Austra. J. Eco. 22, 298-306 27 Swabrick et al. (1998) Lantana camara L. In: Biology of Australian Weeds 2, 119- 140 28 Van Oosterhout et al. (2004) Lantana control manual. Current management and control. 29 Achhireddy N.R. and Singh M. (1984) Allelopathic effects of lantana (Lantnana camara ) on milkweed vine. Weed Sci. 32, 757-761 30 Ahmed et al. 2007 31 Inderjit and R. M. Callaway. (2003) Experimental designs for the study of allelopathy. Plant and Soil 256,1-11. 32 Kruse, M. et al. (2000). Ecological Effects of Allelopathic Plants – a Review. National Environmental Research Institute, Silkeborg, Denmark. 66 pp. – NERI Technical Report No. 315 33 Inderjit and Weiner J. (2001) Plant allelochemical interference or soil chemical ecology? Persp. Plant Eco. Evo. Sys. 4, 3-12. 34 Pollock, J. L. et al. (2008) Allelopathy and plant invasions: traditional, congeneric, and bio-geographical approaches. Bio. Inv. 10, 875-890. 35 Lankau, R. A., et al. (2011) Plant-soil feedbacks contribute to an intransitive competitive network that promotes both genetic and species diversity. J. of Eco. 99, 176-185 36 Popma and Bongers (1988) The effect of canopy gaps on growth and morphology of seedling rainforest species Oec. 75, 625-632