PROSPECTS FOR CLASSICAL BIOLOGICAL CONTROL OF THE AQUATIC INVASIVE WEED POLYSPERMA ()

By

ABHISHEK MUKHERJEE

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2011

1

© 2011 Abhishek Mukherjee

2

To my beloved parents and sister I could not have made it without your inspiration and support

3

ACKNOWLEDGMENTS

I want to express my sincere thanks to my major advisor, Dr. James P. Cuda for giving me the opportunity to work in this project. His constant supervision and constructive criticisms helped me immensely throughout the period of my PhD research.

I also want to express my sincere appreciation to my co-advisor Dr. William Overholt.

His suggestions and guidance were very helpful in improving my dissertation projects.

I owe my sincere thanks to the other members of my advisory committee, Dr.

William T. Haller (Centre of Aquatic and Invasive , UF), Dr. Matthew A.

Gitzendanner (Department of Biology, UF) and Dr. Gregory Kiker (Department of

Agricultural and Biological Engineering, UF) for their support and guidance during my

PhD dissertation research.

I take this opportunity to express my gratitude to Dr. Carol Ellison (CABI Europe,

UK), Dr. K. C. Pujari (Assam Agriculture University, Assam, ) and Dr. Matiyar R.

Khan (Bidhan Chandra Kirshi Viswavidyalaya, West Bengal, India) for their help during surveys in India. In particular, the training I received from Dr. Ellison during field surveys have been a valuable learning experience. I would like to thank Dr. James W. Jones

(Department of Agricultural and Biological Engineering, UF), Dr. Jason K. Blackburn

(Immerging Pathogen Institute, UF) and Dr. William T. Crow (Entomology and

Nematology Department, UF) for participating in collaborative research with me. In addition, I appreciate Dr. Rodrigo Diaz, Matthew Thom (Entomology and Nematology

Department, UF) and Subhadip Pal (Department of Statistics, UF) for their help and encouragements during collaborative research works.

I sincerely acknowledge the help received from the following persons regarding identifications of ; Lyle Buss, Drs. Paul M. Choate and Howard J. Frank

4

(Entomology and Nematology Department, UF); Drs. Susan E. Halbert, Michel C.

Thomas, Gary J. Steck and Paul E. Skelley (Museum of Entomology, FSCA, DPI,

Gainesville, FL); Dr. Alexander Konstantinov (Systematic Entomology Laboratory,

USDA-ARS); Dr. Laura T. Miller (West Department of Agriculture); Dr. Eric

Guilbert (Muséum National d'Histoire Naturelle, Département de Systématique et

Evolution, France); Dr. Daniel J. Bickel (Australian Museum, NSW, ); Dr.

Andrew Short (University of ); Dr. Zbyněk Kejval (Muzeum Chodska, Czech

Republic); Dr. Charles L Bellamy (California Department of Food and Agriculture) and

Dr. Shen-Horn Yen (National Sun Yat-Sen University, Taiwan).

Words fail to express my thanks to Daniel Okine for his immense help throughout my PhD research. I also want to express thanks to Judy Gillmore and Dr. Julio Medal.

Thanks to all my lab mates, Onour E. Moeri, Karen Stratman and especially, Lindsey R.

Christ for their great support.

Special thanks to my very good friends, Dr. Monohar Chakraborty (Ohio State

University), Rajtilak Majumdar (University of New Hampshire) and Triparna Lahari

(University of Manitoba) for their inspiration and support.

Finally, I breathe my deepest revere to my parents, loving sister and other relations. Acknowledgement is not enough for their unfading sacrifices, love and warmth, which nourished my hopes and ambitions.

5

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 9

LIST OF FIGURES ...... 10

ABSTRACT ...... 12

CHAPTER

1 INTRODUCTION ...... 14

Review of Literature ...... 15 Native Range ...... 15 Description ...... 16 Growth Habit ...... 17 Invasion History ...... 17 Impacts and Threats ...... 19 Need for Research ...... 20

2 MICROSATELLITE AND CHLOROPLAST DNA DIVERSITIES OF IN ITS NATIVE AND INVASIVE RANGES ...... 24

Methods ...... 27 Collection and DNA Extraction ...... 27 Microsatellite DNA Typing ...... 27 Chloroplast DNA Typing ...... 30 Results ...... 31 Microsatellite DNA Typing ...... 31 Chloroplast DNA Typing ...... 32 Discussion ...... 33

3 EFFECT OF SIMULATED HERBIVORY ON GROWTH OF THE INVASIVE WEED HYGROPHILA POLYSPERMA: EXPERIMENTAL AND PREDICTIVE APPROACHES ...... 48

Materials and Methods...... 50 Experimental Set Up ...... 50 Data Collection ...... 51 Model Description ...... 51 Model Simulation ...... 54 Test of predictive accuracy ...... 55 Test of model usefulness ...... 55

6

Statistical Analysis ...... 56 Mesocosm experiment ...... 56 Predictive model ...... 57 Results ...... 59 Mesocosm Experiment ...... 59 Stem length and stem dry weight ...... 59 Root length and root dry weight ...... 60 area and leaf dry weight ...... 60 Final biomass ...... 60 Relative growth rate (RGR) ...... 61 Model Simulation Results ...... 61 Test of prediction accuracy ...... 61 Test of usefulness ...... 63 Discussion ...... 63 Mesocosm Experiment ...... 63 Mathematical Modeling ...... 65

4 EXPLORATORY SURVEYS IN NATIVE AND INVASIVE HABITATS TO IDENTIFY NATURAL ENEMIES ASSOCIATED WITH HYGROPHILA ...... 74

Materials and Methods...... 77 Surveys in Florida ...... 77 Catalog and Geoposition of Herbaria Records ...... 77 Exploratory Field Surveys in Hygrophila‟s Native Range ...... 78 Results and Discussion...... 79 Surveys in Florida ...... 79 Catalog and Geoposition of Indian Herbaria Records ...... 80 Exploratory Field Surveys in Native Range ...... 80 Coleoptera ...... 81 ...... 83 ...... 84 Pathogen...... 87 Conclusion ...... 89

5 PHYTOPARASITIC NEMATODES ASSOCIATED WITH THE RHIZOSPHERE OF HYGROPHILA ...... 103

Materials and Methods...... 103 Sampling and Enumeration of Nematodes ...... 103 Assessment of Nematode Diversity ...... 104 Results ...... 106 Nematode Diversity ...... 106 Phytoparasitic Nematodes Recorded ...... 107 Discussion ...... 108

6 PRIORITIZING AREAS IN THE NATIVE RANGE OF HYGROPHILA FOR SURVEYS TO COLLECT BIOLOGICAL CONTROL AGENTS ...... 118

7

Materials and Methods...... 120 Species Occurrence Data ...... 120 Environmental Data Layers ...... 122 Ecological Niche Modeling ...... 123 Statistical Validation of Model Accuracy ...... 124 Results and Discussion...... 127 Statistical Tests Results ...... 127 Native Range Predictions ...... 128 Conclusion ...... 131

7 SUMMARY ...... 139

APPENDIX

A RECORDS OF HYGROPHILA POLYSPERMA COLLECTED FROM THE CENTRAL NATIONAL HERBARIUM, HOWRAH, INDIA ...... 144

B ROYAL BOTANIC GARDENS, KEW, HERBARIUM RECORDS FOR HYGROPHYLIA POLYSPERMA IN THE OLD WORLD ...... 150

C DESCRIPTIONS OF FIELD COLLECTION SITES IN INDIA ...... 153

D DESCRIPTIONS OF FIELD COLLECTION SITES IN BANGLADESH ...... 156

LIST OF REFERENCES ...... 158

BIOGRAPHICAL SKETCH ...... 179

8

LIST OF TABLES

Table page

1-1 Hygrophila species widespread in Indian subcontinent ...... 23

2-1 Chloroplast DNA regions used to characterize native and invasive populations of hygrophila ...... 39

2-2 Microsatellite loci and primer sequences used in this study ...... 40

2-3 Allelic diversities of microsatellite loci across invasive and native populations of hygrophila ...... 41

3-1 Variables and parameters used in the plant growth model ...... 68

3-2 Results of the statistical analyses used to evaluate model performance ...... 69

4-1 Insects collected from hygrophila during surveys in Florida ...... 91

4-2 Insects collected from hygrophila during surveys in India and Bangladesh ...... 92

4-3 Comparison of aecial rust collected from Hygrophila polysperma with that collected from H. phlomoides, H. salicifolia and H. spinosa ...... 94

5-1 Phytoparasitic nematode species recorded from the root zone of hygrophila in India ...... 110

5-2 Phytoparasitic nematode genera recorded from the root zone of hygrophila in Florida...... 111

6-1 Result of threshold dependent binomial tests of omission and area under the curve for predictions with invasive occurrences...... 133

6-2 Result of threshold dependent binomial tests of omission and area under the curve for predictions with native occurrences ...... 134

9

LIST OF FIGURES

Figure page

2-1 Map showing collection sites of plant samples across invasive and native ranges of hygrophila ...... 43

2-2 Inference of the number of unique genetic clusters for invasive and native populations of hygrophila...... 44

2-3 Assignment of hygrophila individuals collected from invasive and native habitats into district clusters ...... 45

2-4 Inference of the number of unique genetic clusters for native populations of hygrophila...... 46

2-5 Assignment of hygrophila individuals collected only from native habitats into district clusters ...... 47

3-1 Effects of defoliation treatments on plant growth parameters of hygrophila...... 70

3-2 Mean observed vs. predicted final biomass, leaf dry weight, leaf area and axis dry weight summarized across all treatments...... 71

3-3 Specific leaf weight and partitioning coefficient of plants subjected to different levels of defoliation...... 72

3-4 Weekly progression of observed and predicted biomass and average relative growth rate...... 73

4-1 Survey sites in Florida to document natural enemies associated with hygrophila in its invasive habitat ...... 95

4-2 Distribution of hygrophila in India based on herbaria records collected from the Central National Herbarium...... 96

4-3 Survey sites in India and Bangladesh...... 97

4-4 Trachys sp. (Coleopter: ) collected from H. polysperma ...... 98

4-5 and adult of bilinealis (Lepidoptera: ) collected from H. polysperma...... 99

4-6 Nodaria sp (Lepidoptera: ) collected from H. polysperma...... 100

4-7 Precis almana (Lepidoptera: Nymphalidae) collected from H. polysperma ...... 101

4-8 Rust fungus (Puccinia sp.) collected from H. polysperma...... 102

10

5-1 Survey sites in India ...... 112

5-2 Survey sites in Florida ...... 113

5-3 Number of species, density, Shannon diversity and evenness calculated across sampling sites in India...... 114

5-4 Number of , density, Shannon diversity and evenness calculated across sampling sites in Florida...... 115

5-5 Cluster analysis of phytoparasitic nematode assemblage in India...... 116

5-6 Cluster analysis of phytoparasitic nematode assemblage in Florida ...... 117

6-1 Occurrences of hygrophila in the US and Mexico...... 135

6-2 Frequency histogram of probabilities received by known native occurrences of hygrophila...... 136

6-3 Projected distribution of hygrophila in India...... 137

6-4 Projected distribution of hygrophila in Bangladesh...... 138

11

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy

PROSPECTS FOR CLASSICAL BIOLOGICAL CONTROL OF THE AQUATIC INVASIVE WEED HYGROPHILA POLYSPERMA (ACANTHACEAE)

By

Abhishek Mukherjee

August 2011

Chair: James P. Cuda Major: Entomology and Nematology

Hygrophila, Hygrophila polysperma (Roxb.) T. Anders, is an invasive aquatic weed in the southeastern . In this study, I evaluated the prospects for classical biological control of this invasive weed. The objectives of my study were to (i) examine the genetic variability of hygrophila, (ii) study its response to artificial defoliation, (iii) conduct surveys in the native range to identify prospective natural enemies, (iv) study the diversity of phytoparasitic nematodes in the rhizosphere of hygrophila, and (v) develop niche based prediction models to prioritize areas in the native range for future surveys.

The genetic characterization study showed no phylogenetically informative variations at four chloroplast DNA regions. However, microsatellite DNA indicated substantial genetic variation in native populations on hygrophila. The invasive populations were genotypically nearly identical, suggesting that invasive populations may have originated form a single source.

The herbivory simulation study showed that defoliation significantly affected growth and biomass accumulation of this weed. An empirical plant growth model also

12

was successfully developed for a priori understanding of response of hygrophila to different levels of defoliation.

Explorations for natural enemies were conducted in the invasive (Florida) and native (India and Bangladesh) ranges of hygrophila. In Florida, hygrophila experienced very little herbivore damage. A number of insects, including two caterpillars (Precis almana L., Lepidoptera: Nymphalidae and Nodaria sp., Lepidoptera: Noctuidae) that defoliates emerged plants, an aquatic caterpillar (Parapoynx bilinealis Snellen,

Lepidoptera: Crambidae) that feeds on submerged hygrophila, and a leaf mining

(Trachys sp., Coleoptera: Buprestidae), were collected during surveys in the native range. In addition, a very damaging aecial rust fungus (Puccinia sp.) also was collected.

Phytoparasitic nematodes were found to be associated with the rhizosphere of hygrophila in its native (seven genera) and exotic (10 genera) habitats. In comparison to

Florida, significantly higher densities of phytoparasitic nematodes were found to be associated with hygrophila in India.

A niche modeling study to prioritize areas in the native range for future surveys predicted the presence of hygrophila in the north-eastern regions of India and most parts of Bangladesh. Based on the results of this study, new areas were prioritized for future surveys in India and Bangladesh.

13

CHAPTER 1 INTRODUCTION

Amid changing climatic conditions, invasive species have globally become an increasingly recurrent problem. According to Pimentel et al. (2000; 2005), approximately

50,000 non indigenous species have been introduced to the United States (US) accounting for an estimated annual loss of $120 billion in agriculture, forestry and other segments of society. In the US, Florida and Hawaii have proven to be the most vulnerable to exotic species invasions with over 4000 and 925 introduced species, respectively (Cox, 1999).

Due to lack of biotic resistance from Florida‟s depauperate native flora, geographic and climatic suitability, as well as being a major hub for aquatic and foliage industries,

Florida historically has been vulnerable to invasion by non-native flora and fauna (see

Simberloff et al., 1997 for a detailed discussion). The aquatic weed hygrophila,

Hygrophila polysperma (Roxb.) T. Anderson (Acanthaceae), is one such plant that is invasive in the warm water bodies and shoreline in the southern United States. It is a federal listed noxious weed (USDA, 2006), a Florida state listed category II prohibited plant (FLDEP, 2007) and a Florida Exotic Pest Plant Council category I invasive species

(FLEPPC, 2009). It causes serious problems by displacing more desirable aquatic vegetation (Spencer and Bowes, 1985; Sutton, 1995). In addition to the United States, hygrophila is invasive in Australia where it has been declared as a state listed Class 1 noxious weed in New South Wales and placed on the restricted list in Western Australia

(Australian Weeds Committee, 2010). In addition, it also is present in Mexico

(Kasselmann, 1994; Mora-Olivo et al., 2008) and North Rhine-Westphalia state of

Germany (Hussner et al., 2007; Hussner et al., 2010).

14

Hygrophila belongs to the family Acanthaceae, subfamily Ruellioideae, tribe

Ruellieae, and subtribe Hygrophilinae (Long, 1970). The tribe Ruellieae includes the genera Blechum P. Brown, Dyschoriste Nees, Hygrophila (Roxb.) T. Anders and Ruellia

L. (Cook and Cook, 1996). The genus Hygrophila includes the subgenera Asteracantha and Hygrophila. About 148 species of Hygrophila have been recorded worldwide (GBIF,

2011).

The center of origin of the genus Hygrophila is the Old World tropics, including

Indochina and Malaysia with only a few African and American species (Long, 1970).

Many Hygrophila species show high degree of gross morphological similarity, requiring the use of epidermal features for species level identifications (Ahmad, 1976). Apart from

H. polysperma, seven other species of Hygrophila are widespread in the Indian subcontinent (Table 1-1) (Cook and Cook, 1996).

In the US, beside H. polysperma, three non-native species of Hygrophila are available commercially as decorative aquarium plants. These are giant hygro

(Hygrophila angustifolia R. Brown), temple plant (H. corymbosa Lindau) and water wisteria (H. difformis Blume) (Coile, 1995). The gulf swampweed,

Nees et al. (= H. lacustris, = H. brasiliensis) is the only known native North American representative of the genus Hygrophila (Wunderlin and Hansen, 2003).

Review of Literature

Native Range

Hygrophila is commonly known as hygro, East Indian hygro, green hygro, Miramar weed, oriental ludwigia and Indian swampweed. polysperma Roxb.,

Hemidelphis polysperma var. joshianus Rao et Biswas (Cook 1996) and Hemidelphis polysperma (Roxb.) Nees are its junior synonyms. It is an Old World species and is

15

believed to be native to India (Rataj and Horeman, 1977) or broadly to the Southeastern

Asiatic mainland (Les and Wunderlin, 1981; Spencer and Bowes, 1985; Angerstein and

Lemke, 1994). According to USDA (2010), the native range of hygrophila includes

Bangladesh, Bhutan, Cambodia, , India, Laos, Myanmar, Nepal, Thailand and

Vietnam (USDA, 2010). Cook (1996) included Pakistan in the native range of this plant.

In India, hygrophila is present in almost all states between 0 - 1200 m of mean sea level, including, Andhra Pradesh, Assam, Bihar, Delhi, Gujarat, Karnataka,

Maharashtra, Madhya Pradesh, Punjab, Rajasthan, Uttar Pradesh and West Bengal

(Cook 1996).

Plant Description

Hygrophila is a perennial terrestrial or riparian weed, up to ~1.5 m tall with sparsely hairy squarish stems that are ascending or creeping with abundant elongate or rarely rounded cystoliths (deposits of calcium carbonate) in the epidermis (Angerstein and Lemke, 1994). It is usually rooted in the substrate, but when submerged, hygrophila also roots freely at the floating nodes. The of hygrophila are opposite, broadly elliptic to oblanceolate, minutely denticulate or entire, up to 8cm long and up to 2cm wide, usually broader toward the tip, and are sessile with the bases joined at the nodes by ciliated flanges of tissue. Both abaxial and adaxial surfaces of leaves contain cystoliths (Angerstein and Lemke, 1994). Although hygrophila does not show heterophylly, submerged leaves are generally larger in size and lighter in weight than emersed leaves (Sutton, 1995). Heterophylly is a condition shown by some aquatic plants, where depending on growth habitat (submerged or emergent), the leaf shape differs significantly. Solitary sessile develop at the upper leaf axils of the emergent stems and are surrounded by two hairy modified, narrowly lanceolate 16

herbaceous bracts (~4 – 5 mm). Sepals are united basally, green, hispid and equally 5 - lobed. The corolla is bluish white, hairy ~5 - 6 mm long, with two lobbed upper lip and three lobbed lower lip. The is characterized by two with two celled anthers having glabrous filaments. The ovary is distally hispid with a long style (~3mm) and flattened stigma (~0.3 mm). The fruit is a narrow capsule, ~9mm long covered with hairs especially near the top that splits longitudinally to release tiny flattened, round,

~0.8 mm diameter seeds (Angerstein and Lemke, 1994).

Growth Habit

Hygrophila can grow as a rooted submerged or emerged plant in shallow water and on saturated shorelines (Sutton, 1996). In south Florida, hygrophila can grow perennially with maximum production of biomass recorded during summer and early fall

(June - October) (Sutton, 1995; Cuda and Sutton, 2000). Nutrients of sediment, day length and water temperature affect growth rate of this weed (Cuda and Sutton, 2000).

In laboratory tests, the optimum water temperature for photosynthesis was found to be

25oC. However, rate of photosynthesis was recorded to be adequate for growth of this weed between 10o to 35oC, suggesting that water temperature in Florida will not restrict the growth of this invasive species (Spencer and Bowes, 1985). Flowers developed during late October to early March. A high percentage of seed set has been recorded in

Florida (Sutton, 1995). Mature seed capsules develop between December and February

(Schmitz and Nall, 1984).

Invasion History

Hygrophila was first imported into Ohio by an aquarium nursery at the end of

World War II (Innes, 1947). Although, initially known as Oriental Ludwigia, it was later identified as Hygrophila polysperma T. Anders at the University of Pennsylvania

17

(Reams, 1953; Schmitz, 1985). In the US, hygrophila is well established in Florida and

Texas and possibly in Kentucky and Virginia (EDDMaps, 2010).

In Florida, hygrophila was first collected from Tampa in 1965. Until 1977, it was misidentified as Dyschoriste sp. Nad and was later correctly identified by Dieter C.

Wasshausen at the Smithsonian Institute (Les and Wunderlin, 1981). During the 1970s, hygrophila was found infesting the Able Canal draining the Caloosahatchee River in the

Lee County, Florida. On the east coast of Florida, infestations were also recorded in

Miramar and City of Margate canals, part of the Everglades drainage system (Les and

Wunderlin, 1981). Available records indicated that during the 1980s, the area infested by hygrophila increased approximately ten-fold, including notable infestations in the

Loxahatchee River in 1986 and Withlacoochee River in 1989. Northward expansions of habitats infested by this weed also occurred during that time with an infestation recorded at the Santa Fe River in Columbia County. By 1999, hygrophila was known from at least 22 rivers/streams, 13 lakes, 2 ditches and 7 canal systems in Florida

(Cuda and Sutton, 2000). According to the Florida Department of Environment

Protection (FLDEP), Annual Report for Fiscal Year 2006 - 2007, hygrophila was reported from 25 water bodies covering ~254 acres (FLDEP, 2007).

Hygrophila was most likely introduced into river systems either directly through cultivation by local aquatic plant nurseries or indirectly through careless dumping by aquarists (Angerstein and Lemke, 1994). It was first reported from the San

Marcos River in Hays County during 1960 (Angerstein and Lemke, 1994). During the

1970s additional areas of invasion were reported, including Sessoms Creek in Hays

County (Angerstein and Lamke 1994). In 1994, hygrophila was discovered in the spring

18

fed portions of the Comal River system, Comal County and in 1998, at the San Felipe springs, located in the far western drainage of Val Verde County (Jacono, 2003).

Impacts and Threats

Hygrophila is an herbaceous perennial plant. It is capable of forming dense stands which can occupy the entire water column, causing disruption in irrigation and flood control systems (Schmitz and Nall, 1984; Sutton, 1995). Owing to the fast growth and rapid spread of hygrophila, it has the potential to shade out and outcompete native submerged plants in shallow water (Spencer and Bowes, 1985) and other lotic habitats

(Angerstein and Lemke, 1994). It also can create problems in shore line areas especially in rice fields (Krombholz, 1996).

In addition to its aquatic habit, hygrophila often grows terrestrially along the shore as a ditch bank weed (Spencer and Bowes, 1985). The plant has a high growth rate and is capable of rapidly expanding the population ten-fold in a year (Vandiver, 1980).

Rooted nodes of small pieces of the easily fragmented stems have the potential to develop new stands (Les and Wunderlin, 1981). Spencer and Bowes (1985) showed that regrowth potential from stem fragments even surpasses that of Hydrilla verticillata (L.f.) Royle (Hydrocharitaceae). In a recent experiment, Doyle et al. (2003) reported that Ludwigia repens J.R. Forst. (Onagraceae) is susceptible to displacement by hygrophila due to its higher bio-mass production potential. Similarly, study by Owens et al (2001) highlighted that hygrophila can replace the endangered Texas wild rice

(Zizania texana Hitchc., ) in the San Marcos river system in Texas.

The lack of seasonal variation in biomass, low light compensation and saturation points, a low CO2 compensation point, and the ability to rapidly change resource acquisition in response to changing environmental conditions make hygrophila a good 19

competitor (Spencer and Bowes, 1985; Kovach et al., 1992). Schmitz (1985) proposed several reasons why hygrophila is causing problems in Florida: (i) the plant has become adapted to Florida water ways; (ii) it was growing in out of the way places and only now is introduced into public water bodies; (iii) the nutrient level in Florida‟s water ways may have reached the critical limit due to urbanization, stimulating the growth of this plant;

(iv) its relative tolerance to common aquatic herbicides; and (v) failure of plant- managers to identify this plant initially due to its similarity with alligatorweed

( phylloxeroides Grisebach, Amaranthaceae).

Need for Research

Practical solutions for long term control of this weed are not readily available. The existing control measures are not only costly but also do not provide adequate control.

Mechanical control may be useful for removing the floating mats, but the action of mechanical harvesters increases the number of stem fragments that can be transported to other areas and thus hastens the spread of hygrophila (Sutton, 1995).

Registered aquatic herbicides do not provide effective control of this invasive weed

(Vandiver, 1980) and can be expensive. For example, endothall provides temporary control of both submerged and emerged forms but regrowth occurs within 4 to 8 weeks post-treatment (Sutton, 1995). Multiple applications of endothall are required to keep hygrophila under control (Sutton, 1995). Tank mixes combining copper with diquat showed little effect on this plant in Florida canals 4 weeks after application (Sutton,

1995). Herbicides typically used for controlling hygrophila also are expensive, costing from $988 to $1482 per hectare. Treatment costs are even higher when labor and equipment are included. An extreme case involved the use of fluridone in a flowing water system where sustained control was achieved for 20 months at a cost of $34,580 20

per hectare (Sutton, 1996). Chemical treatments for controlling hydrilla, may even contribute to the spread of hygrophila by leaving open areas that are susceptible to invasion by this plant (Schmitz and Nall, 1984; Spencer and Bowes, 1985; Sutton,

1995).

The herbivorous Chinese grass carp (Ctenopharyngodon idella Valenciennes) is reported to have low preference for hygrophila (Cassani, 1996). Grass carp usually feed on submerged hygrophila to a limited extent in the absence of preferred food plants

(Sutton, 1995). The epidermal cell walls of hygrophila stems and leaves contain cystoliths (Sutton, 1995). The presence of cystoliths renders this plant less palatable to grass carp (Sutton and Vandiver Jr, 2006). More importantly, because grass carp are not selective feeders; they will completely remove all aquatic vegetation in a water body if densities are not precisely controlled.

Hygrophila is a threat to all Florida fresh water bodies because it is capable of tolerating a wide range of water temperature and the seeds or viable fragments it produces can be transported unintentionally to new locations. Les and Wunderlin (1981) believed that owing to its very high reproductive potential and demonstrated vigor in

Florida water, hygrophila could eventually become another noxious weed. The weed potential of hygrophila appeared to be very high due to its ability to grow as both a submerged and an emergent plant, ability for high biomass production and development of a dense canopy at the water surface (Doyle et al., 2003). Thus, a combination of factors suggests that in the absence of effective control methods, range expansion of hygrophila will continue. Recent experiences in south Florida indicate that practical solutions for long term control of this plant are not currently available. Alternative

21

methods are needed to address the hygrophila problem in Florida in order to prevent the rapid regrowth and spread of these plants. Considering several biological and economic attributes of this weed, Cuda and Sutton (2000) reported that biological control may be a viable option for managing hygrophila. The aquatic habitats infested with hygrophila are relatively stable ecosystems and thus are favorable to biological control agent establishment (Cuda and Sutton, 2000). The specific objectives of this study were, to (i) characterize the genetic variation in exotic and native populations of hygrophila using microsatellite and chloroplast DNA diversities, (ii) study the effect of simulated herbivory on growth of hygrophila and develop a empirical plant growth model

(iii) conduct exploratory surveys in native and invasive habitats to identify candidate natural enemies for biological control, (iv) study the association of phytoparasitic nematodes with rhizosphere of hygrophila in native and exotic habitats, and (v) develop niche based prediction models to prioritize areas in the native range of hygrophila to conduct future surveys

22

Table 1-1. Hygrophila species (beside H. polysperma) widespread in Indian subcontinent

Name Species Characteristics

H. balsamica (L.f.) Rafinesque Annual or perennial; found in shallow water H. difformis (L.f.) Blume Perennial, usually found in shallow water, may be bottom-rooted or straggling over surface H. heinei Sreemadhavan Annual, found in shallow water, swamps and rice fields H. pinnatifida (Dalzell) Poorly known, probably found in shallow water Sreemadhavan H. quadrivalvis (F. Hamilton) Nees Perennial or annual; found in shallow water, marsh and along the banks of streams H. schulli (F. Hamilton) M. R. et S. Perennial; found in swamps, temporary pools, M. Almeida edge of the tanks, canals, ditches and in rice field H. serpylla (Nees) T. Anderson Perennial; grows in marshy land, drains, temporary pools and slowly moving water in streams and pools

23

CHAPTER 2 MICROSATELLITE AND CHLOROPLAST DNA DIVERSITIES OF HYGROPHILA POLYSPERMA IN ITS NATIVE AND INVASIVE RANGES

Hygrophila, Hygrophila polysperma (Roxb.) T. Anders (Acanthaceae), is an invasive aquatic weed in the southeastern United States (Cuda and Sutton, 2000), as well as in Mexico (Mora-Olivo et al., 2008) and Australia (Australian Weeds Committee,

2010). Recently, it also was declared as an invasive weed in Germany (Hussner et al.,

2007). In the US, hygrophila from dense stands that can occupy the entire water column, clogging irrigation, navigation and flood control structures (Schmitz, 1985;

Sutton, 1995). In addition, field and laboratory studies have shown that this weed can replace native macrophytes in shallow water with a pH ranging between 5.0 and 7.0

(Spencer and Bowes, 1985; Angerstein and Lemke, 1994). Practical solutions for long term control of hygrophila are not available. The existing control measures are costly and do not provide adequate control (Cuda and Sutton, 2000). Considering several biological and economic attributes of this aquatic weed, Cuda and Sutton (2000) reported that biological control may be a viable approach for managing this weed.

However no information about the natural enemies of this aquatic plant was available until recently (Mukherjee et al., 2008, also see Chapter 4 of this dissertation).

Furthermore, unknown origin(s) of the invasive hygrophila population creates difficulty in prioritizing native region(s) for foreign exploration.

Based on the „Enemy Release Hypothesis‟ (Williams, 1954), classical weed biological control relies on the intentional release of natural enemies to suppress the growth and vigor of the target invasive weed (Keane and Crawley, 2002). To avoid any unwanted consequences, strict host specificity is a crucial requirement for any potential biocontrol agent (McEvoy, 1996). Host specificity between a biocontrol agent and its 24

target weeds is governed by several factors: genetic architecture and evolutionary relationship between the target weed and the biocontrol agent (Schaffner, 2001;

Goolsby et al., 2006), as well as their behavioral (Schaffner, 2001), chemical and ecological relationships (Littlefield and Buckingham, 2004). For example, Goolsby et al.

(2006) demonstrated that a close spatial match between the phytophagous mite,

Floracarus perrepae Knihinicki & Boczek (Eriophyidae) and its host, the invasive Old

World climbing fern Lygodium microphyllum Cav. R.Br. (Pteridophyta: Lygodiaceae) in the native range is a prerequisite for effective control of the invasive fern. In a recent study, Manrique et al (2008) also documented the correlation between Brazilian pepper tree (Schinus terebinthifolius Raddi, Anacardiaceae) genotype and fecundity and longevity of Pseudophilothrips ichini Hood (Thysanoptera: Phlaeothripidae) and another unidentified thrips (Pseudophilothrips sp. near ichini). In recent years, a number of studies have used molecular approaches to compare genetic diversity between invasive and native habitats to reconstruct invasion history and identify source population(s)

(reviewed in Ward et al., 2008). As suggested by Goolsby et al (2006), and also supported by the findings of Manrique et al. (2008), identifying the source population of an invasive weed may help in collecting biological control agents that are closely adapted to the target weed (Croxton et al., 2011). Understanding the role of local adaptation is thought to be crucial for effective management of the target weed (Karban,

1989).

In addition to management implications, study of genetic diversity can generate valuable information regarding demography and invasion history of an invasive weed

(reviewed in Ward et al., 2008). Species invasion is often associated with marked

25

founder effects that significantly deplete genetic diversity of the invading species.

However, a number of recent studies have demonstrated that multiple introductions, along with intra-specific hybridization, can increase the genetic variation of an invasive species. For example, Lavergne and Molofsky (2007) linked the invasion success of the reed canarygrass (Phalaris arundinacea L., Poaceae) in North America to multiple introductions, which increased the genetic diversity, allowing it to invade novel habitats.

In another study, Williams et al (2005) identified two distinct chloroplast DNA haplotypes of the invasive Brazilian peppertree that were introduced separately. Furthermore, they documented that crossing between these two haplotypes increased genetic variation of the invasive weed (Williams et al., 2005) and is thought to be associated with wide climatic tolerance of the invasive haplotypes (Mukherjee et al, unpublished).

As reviewed in Ward et al. (2008) a number of studies have used microsatellite or chloroplast DNA diversity to characterize patterns of genetic variation in populations of an invasive species and also to trace its invasion history, respectively. Microsatellites, also known as simple sequence repeat (SSR) or variable number tandem repeats

(VNTR) are tandem repeats of 1-6 nucleotides found in high frequencies in nuclear genome of most eukaryotes (Salkoe et al. 2006). Microsatellite markers generally have high mutation rates resulting in high allelic diversity. The polymorphic nature of microsatellites provides the high resolution necessary to identity population level genotypic structure of invasive species (Salkoe et al. 2006). On the other hand, chloroplast DNA, being maternally inherited (Mogensen 1996), is frequently used to trace invasion history (Williams et al. 2005), pattern of invasion as well as range expansion (McCauley et al. 2003). As a part the ongoing effort to develop biological

26

control agents for this aquatic weed, I also used a combination of microsatellite markers and chloroplast DNA sequences (i) to characterize the genetic variation in the exotic and native populations of this weed, and (ii) to determine the geographical origin of the invasive hygrophila population.

Methods

Collection and DNA Extraction

In total, 328 individuals were sampled from the native (India and Bangladesh) and invasive ranges (Australia, Mexico, and US) of hygrophila (Fig 2-1). In Florida, 96 individuals were sampled across eight counties (Fig 2-1B). Samples (n = 48) from

Texas were collected from four sites in Hays County. Six plant samples also were collected from the Barkley County, . In Mexico, individuals (n = 20) were collected from two sites in Tamaulipas (Fig 2-1A). In addition, plant samples were collected from two sites from New South Wales, Australia (Fig 2-1C). In total, 124 individuals were collected from the native range, including eight sites from West Bengal and Assam, India and six sites in Mymensingh, Bangladesh (Fig 2-1D). Each sample consisted of silica-gel dried 1-2 cm of the growing tip of hygrophila collected from individual plants approximately 10 m apart. Latitude and longitude of each sampling site were recorded using a GPS. DNA was extracted from plant tissue following a phenol/chloroform protocol modified from Doyle and Doyle (1987) and Cullings (1992).

All DNA extracts were quantified using a NanoDrop® Spectrophotometer (Thermo

Scientific, Wilmington, DE).

Microsatellite DNA Typing

Hygrophila specific microsatellite primers were developed using the next- generation 454 GS-FLX (Roche, Penzberg, Germany) sequence technology. A 454

27

DNA sequencing library was constructed following manufacturer‟s protocols using a

DNA sample extracted from hygrophila collected from Santa Fe River, Columbia

County, Florida. This library was sequenced on a 1/8th plate. A bioinformatics pipeline, modified from Castoe et al. (2010) was used to develop primers, using the following parameters; minimum number of repeat units: ≥ 9, primer distance from microsatellite region: ≥ 30bp.

Microsatellite primers (n = 11, see Table 2-2 for primer sequences) were chosen based on their ability to amplify hygrophila DNA, and generate consistent profiles that could be scored unambiguously. The forward primer of each pair had an M13 tailing sequence added to it (Schuelke, 2000). PCR (10 µl) was run with ~50 ng of DNA, 20µM

MgCl2, 0.5 µM of each dNTP, 0.12 µM of forward primer and 4.5 µM of reverse primer;

4.5 µM of one of four fluorescently labeled M13 primers (6FAM, VIC, NED and PET,

Applied Biosystems), and 0.5 U Taq DNA polymerase. Thermal cycling profile was as follows: one cycle at 94oC (1 min), followed by 30 cycles at 94oC (30 sec), 52 oC (30 sec), and 72 oC (30 sec), followed by a single extension at 72 oC (40 min). Reaction products were separated by capillary electrophoresis with the GeneScan 600 LIZ size standard and ran on an ABI-3730xl Genetic Analyzer (Life Technologies, Carlsbad,

California) at the Interdisciplinary Center for Biotechnology Research, University of

Florida. Genotypes were scored using GeneMarker 1.60 software (SoftGenetics, State

College, PA).

For each microsatellite locus, number of alleles (NA), observed heterozygosity

(HO), Nei‟s gene diversity (also referred to as expected heterozygosity, HE) and inbreeding coefficient (FIS) were calculated across invasive and native populations using

28

GenAlEx 6.41 (Peakall and Smouse, 2006). The allelic richness (AR) and private allele richness (PA) for each locality were calculated (correcting for sampling size) following the rarefaction technique developed by Kalinowski (2004) using HP-RARE (Kalinowski,

2005).

The Bayesian-clustering approach, implemented in the program Structure

(Pritchard et al., 2000) was used to infer the genetic structure of the native and invasive populations of hygrophila. Structure clusters multi-locus genotypic data into K populations (= number of assumed populations) under the assumption that all loci are at linkage equilibrium (LE) and also at Hardy-Weinberg equilibrium (HWE). Under a given ancestry model, an individual multilocus genotype is probabilistically assigned to one or more subpopulations based on an ancestry coefficient (Q). Structure runs were performed using genotype data for all microsatellites (n = 11) under the default ancestry and frequency models (correlated allele frequency model, assuming admixture). The admixture ancestry model assumes that an individual inherited its genome from different subpopulations (Pritchard et al., 2009), a condition usually evident in the case of invasive species (for example, see Ascunce et al., 2011). Ancestry coefficient (Q) calculated by Structure can vary between 0 and 1.0, with 1.0 indicating full membership of an individual to a subpopulation. Following Williams et al. (2005), individuals with Q ≥

0.90 were considered to have pure ancestry to a subpopulation and those with Q < 0.90 were considered as hybrids. Structure simulations were run with 105 burn-in iterations, followed by 106 Marcov chain Monte Carlo iterations. For each K, simulations were run with 10 replications to ensure consistency in the calculations of Q. Selection of optimum number of distinct clusters was based on both the K value at which the likelihood

29

distribution plateaued or deceased (Ascune et al. 2011) and the peak value ΔK statistic

(the second order rate of change of the log-probability of the data) of Evanno et al.

(2005) using a web based application, Structure Harvester

(http://taylor0.biology.ucla.edu/struct_harvest). Each replicated simulation (n = 10) in

Structure analysis generated a distinct membership matrix (Jakobsson and Rosenberg,

2007). The program CLUMPP v 1.1.2 was used to generate a single permuted Q matrix across replicates (Jakobsson and Rosenberg, 2007). Distributions of Q across populations were plotted using the program DISTRUCT v 1.1 (Rosenberg, 2004).

Structure analyses were conducted with three partitions of the data: (i) all sampled individuals from the native and invasive ranges (K = 1 to 5), (ii) individuals only from the invasive habitats (K = 1 to 4) and (ii) individuals only form the native populations (K = 1 to 15). The choice of K was dependent on the number of distinct population from which hygrophila was collected. For example, in the first analysis, the maximum K was set to five, as individuals were collected from five countries (Australia, Bangladesh, India,

Mexico and US). In case of analysis with all native samples, maximum K was set to 15

(14 spatially unique populations from where hygrophila samples were collected, Fig 2-1, plus one) to ensure that Structure can identify the distinct genotypic cluster present within the population.

Chloroplast DNA Typing

Invasive and native samples were sequenced at four chloroplast DNA regions: the trnS-G spacer (Hamilton, 1999), psbM-trnD spacer, rpl16 gene (Shaw et al., 2005) and trnL-F spacer (using "c" and "f" universal primers, Taberlet et al., 1991). Table 2-1 summarizes the sequences of the forward and the reverse primers and polymerase chain reaction (PCR) conditions used in this study. PCR products were sequenced form 30

both ends on an ABI 3730xl Genetic Analyzer (Life Technologies, Carlsbad, California) using the same primers used for amplifications of respective regions. In total, I sequenced 48 individuals across invasive and native populations of hygrophila with at least one sample per population. Sequences were viewed, trimmed and contiged using

Sequencher 4.2 (GeneCodes Corporation, Ann Arbor, Michigan) and were aligned manually using BioEdit (Hall, 1999). PAUP* v 4b10 (Swofford, 1999) was used for phylogenetic analysis of sequence variations.

Results

Microsatellite DNA Typing

All the microsatellite loci were polymorphic in the native populations (Table 2-3).

The number of alleles (NA) across loci ranged between 2 and 10, with an average NA of

4.9 for populations collected from West Bengal and Assam in India and 5.8 for those collected from Mymensingh, Bangladesh. In general, observed heterozygosities across native populations were significantly higher than expected. Average FIS’s observed for

Indian and Bangladesh populations were 0.5 and 0.3 respectively (Table 2-3). Average

AR observed was 4.03 and 4.04 for Indian and Bangladesh populations, respectively.

In contrast, in the invasive range, heterozygosity was observed only for three loci across all invasive populations, Hyg05AT13, Hyg07AT11, and Hyg10ATT15. Two other loci, Hyg19AT11 and Hyg25ATT14 were heterozygous only for Mexican and Florida populations, respectively (Table 2-3). All heterozygous loci had significant heterozygote excess (FIS range: -0.67 to -1.0) with no homozygotes being observed in a population at polymorphic loci, suggesting that invasive hygrophila populations propagate primarily via asexual reproduction. The mean allelic richness (AR) ranged between 1.27 and 1.35.

31

Average private allele richness (PA) was low (range: 0 to 0.03), with no PA found for

Australian population or populations from Texas and South Carolina in the US.

The Bayesian clustering analysis implemented with all microsatellite genotypes indicated that all individuals were assigned with high probability (Q ≥ 0.90) to one of two distinct clusters (K = 2), native and invasive (Fig. 2-3). When plotted against K, both the mean estimated Ln-likelihood of data conditional on K [ln Pr(X|K)] and ΔK statistic also inferred that the K = 2 is the optimum number of cluster (Fig. 2-2). No hybrid individuals between invasive and native clusters (Q < 0.90) were observed.

Structure analyses for invasive habitats (K = 1 to 5) demonstrated that the invasive population cannot be separated into separate clusters, strongly suggesting all sampled individuals from the invasive habitats were from a single origin. Cluster analyses of individuals from the native range generated ambiguous results regarding optimum number of K. When plotted, the highest log-likelihood value of the data occurred at K =

5. On the contrary, peak value of the ΔK statistic occurred at K = 2 (Fig. 2-4). Figure 2-5 summarizes the results for K = 2 to 5. When compared, K = 5 clusters generated the maximum number of unique clusters, that reflects a biologically meaningful pattern of the native populations of hygrophila.

Chloroplast DNA Typing

Very low sequence variations were observed across invasive and native populations for all the chloroplast DNA regions used in this study. Results of the PAUP indicated that none of the variations observed during sequence alignment, were phylogenetically informative and therefore cannot be used to infer genetic architecture of hygrophila.

32

Discussion

The Structure simulations with microsatellite data from both invasive and native ranges clearly demonstrated that the invasive populations of hygrophila are genotypically nearly identical, suggesting that the invasive population most likely originated from a single source. Considering the history of invasion of hygrophila, this result suggests that US populations, where the species was first documented in 1947, could be the source for introductions to other invasive habitats (Australia and Mexico).

The earliest record of hygrophila in the US was an importation into Ohio by an aquarium nursery at the end of World War II (Innes, 1947). It is currently well established in Florida and Texas and has been recorded from Alabama and South

Carolina (EDDMaps, 2010). In Florida it was first collected from natural areas in Tampa in 1965. Herbaria records of hygrophila from Texas indicated that it was first collected from natural areas in 1965 (summarised in Angerstein and Lemke, 1994). In contrast, the invasions into Alabama and South Carolina were more recent and were first documented in 2010 (EDDMaps, 2010). Hygrophila was first reported from Tamaulipas region of Mexico in1997 and it was assumed that the Texas population was the source of this invasion (Mora-Olivo et al., 2008). No definitive information is available on invasion history of this weed in Australia. It was declared as a Class 1 noxious weed in

New South Wales and placed on the restricted list in Western Australia in 2010

(Australian Weeds Committee, 2010).

As mentioned earlier, microsatellite data demonstrated that very little variation exists across invasive populations of hygrophila. Lack of genetic variation across invasive populations can be explained under three scenarios. First, if there is limited variation in the native range, all introduced populations could be identical by chance. 33

Second, three independent introductions from same native source population can lead to genotypic uniformity across invasive habitats. Lastly, the invasive populations could have arisen through secondary invasions from other introduced populations. In the native range, hygrophila has substantial inter- and intra-population level genetic variations, making the first two hypotheses seem unlikely (Table 2-3). Additionally, intra- population variation is great enough that even if it is assumed that a single population acted as the source, it is highly unlikely that all introduced populations would be genotypically identical.

The most plausible scenario is that this weed was first introduced into the US and this invasive population subsequently acted as the source population for invasion to

Australia and Mexico. This scenario is biologically meaningful and also is supported by the documented history of invasion of hygrophila. Based on available information, hygrophila was first reported as an invasive weed in the US. Based on invasion history and spatial proximity, Mora-Olivo et al. (2008) assumed that the US population was the source of Mexican invasion. As reported in this study, lack of any genetic variation between US and Mexican populations supports this assumption. Although no definite records are available, the genotypic similarity between US and Australian populations suggests that the US may be the source of the Australian population. Hygrophila is an important aquarium plant and is imported worldwide by the aquarium plant and foliage industries (for example, see Brunel, 2009). Similar to the US (Innes, 1947), this plant could also have been imported into Australia as an aquarium plant and later established as an invasive weed.

34

Interestingly, a recent study demonstrated that the invasion history of fire ants

(Solenopsis invicta Buren, Formicidae: Hymenoptera) followed a similar path (Ascunce et al., 2011). In their study, Ascunce et al. (2011) demonstrated that fire ants were introduced into the US from South America and the US population subsequently acted as the source for global invasion of this species. Their hypothesis was supported both from the genotypic (microsatellite and the nuclear gene Gp-9) and haplotypic (variations in mitochondrial DNA sequences) variations recorded across invasive and native fire ant populations (Ascunce et al., 2011). As mentioned earlier, none of the chloroplast DNA regions used in this study generated any phylogenetically meaningful variations to reconstruct the invasion history of hygrophila. Therefore the above stated hypothesis regarding possible origin of this weed is solely based on the multilocus microsatellite diversities recorded across invasive and native ranges of this weed.

Structure analysis with invasive and native populations divided individuals into two distinct clusters, one for invasive populations and other for native populations. No native individual was found to share ancestry with those from invasive regions, indicating that this study failed to identify a putative source of invasive hygrophila populations. This failure can be attributed to insufficient sampling in the native habitat. Given the substantial genetic variation observed across native populations of hygrophila (Table 2-

3), small sample sizes can reduce the chances of finding the genotype that served as the source of the initial introduction. This weed is known to have a wide native range, including all of southeastern Asia and parts of China (Rataj and Horeman, 1977; Cook and Cook, 1996; USDA, 2010). In this study, native samples were collected only from parts of India (Assam and West Bengal) and the Mymensingh region of Bangladesh,

35

representing a very small portion of the native habitat. As emphasized by Murihead et al. (2008), lack of sampling across native populations can significantly affect the ability of population genetic studies to identify the source of introduced populations. In addition, small sample size can affect population level genetic differentiation, a crucial factor for accurate identification of putative source populations (Murihead et al., 2008).

Therefore, detailed studies involving more samples per population and a greater representation of the native range are necessary to identify the source population.

However, it is important to note that the primary purpose of this study was to identify natural enemies of hygrophila, not determine the origin of invasive populations.

Native areas were selected for surveys based on information collected form herbaria records (see Chapter 4 for methods and Appendices A and B for label information collected). Herbaria records indicated that northeastern parts of India, in particular, the state of West Bengal was very favorable for growth of hygrophila (see Chapter 4 for more details). Therefore, survey efforts were concentrated in the eastern parts of India.

Considering the climatic similarities, surveys also were conducted in parts of

Bangladesh.

In addition to insufficient sampling, the differences in primary mode of reproduction between invasive and native populations of hygrophila could also affect the ability to identify the putative source populations. The significant excess in heterozygote frequencies, as observed across invasive habitats, strongly suggests that clonal propagation is the primary mode of reproduction in these populations (Table 2-3). Other studies on invasive hygrophila also provide similar evidence (for example, Les and

Wunderlin, 1981; Sutton, 1995). In contrast, as evident from the substantially positive

36

FIS, self-fertilization is probably the primarily mode of reproduction in the native range of hygrophila. The presence of autogamy in hygrophila has been confirmed by Les and

Wunderlin (1981). Similar evidence also was obtained from a closely related species,

Ruellia succulenta Small (Acanthaceae) (Geiger et al., 2010). Note that chloroplast DNA diversity studies on different species of the family Acanthaceae have confirmed that the two genera, Hygrophila and Ruellia belong to the same clade (McDade and Moody,

1999). Self-fertilization, an extreme case of positive assortative mating is known to deplete the heterozygosity in a population (Lowe et al., 2004). Under such a scenario, an allele present in the historical source population can be lost as a result of loss of heterozygosity under successive generations of self-fertilization. On the other hand, clonal propagation, as seen in the invasive habitats of hygrophila, will fix heterozygosity over generations, leading to allelic divergence from the native habitat.

The difference in mode of reproduction may be attributed to the difference in prevailing ecological conditions of the invasive and native ranges of hygrophila. Based on my field observations, hygrophila in its native range grows primarily as a terrestrial plant (see Appendices C and D for description of native survey sites). In contrast, in its invasive range, it generally grows as a submerged or emergent plant, which favors asexual reproduction (Les and Wunderlin, 1981). Although seed set has been observed in Florida populations of hygrophila (Les and Wunderlin, 1981), a number of studies have confirmed that in the invasive range it has profuse vegetative reproduction

(Spencer and Bowes, 1985; Angerstein and Lemke, 1994; Sutton, 1995). Differences in mode of regeneration based on available habitat are known for other plants. For example, the aquatic plant Butomus umbellatus L. (Butomaceae) reproduces both

37

sexually and asexually thorough the production of propagative bulbils (Eckert et al.,

2003). A similar case has been reported for the aquatic tape grass Vallisneria spp.

(Alismatales) (Lokker et al., 1994). Earlier studies demonstrated that higher genetic diversities are generally observed in clonal terrestrial species that can reproduce sexually (for example, Trifolium repens L., Fabaceae) compared to species that reproduce vegetatively (Ellstrand and Roose, 1987).

Based on the above information, my hypothesis regarding effects of habitat condition and consequently on genetic diversities of invasive and native hygrophila population seems plausible. However, further research will be necessary to generate direct evidence in this regard.

38

Table 2-1. Chloroplast DNA regions used to characterize native and invasive populations of hygrophila

Regions 5‟ – 3‟ Sequences PCR Conditions References trnS - F: GCCGCTTTAGTCCACTCAGC 94oC - 5min, 35x (94oC - 30s, Hamilton trnG 52oC - 45s, 72oC - 45s) 72oC 1999 - 40min R: GAACGAATCACACTTTTACCAC psbM - F: AGCAATAAATGCRAGAATATTTACTTCC 80 oC - 1min, 4x (94oC - 1min, Shaw et al. trnD AT 50oC - 1min + 1oC /cycle, 2005 72oC - 3min 30s), 30x (94oC - R: GGGATTGTAGYTCAATTGGT 1min, 55oC - 1min, 72oC - 3min 30s), 72oC -5min rpL16 F: GCTATGCTTAGTGTGTGACTCGTTG 95oC - 2min, 35x (95oC – 1min, 58oC - 1min 30s, 72oC – 2min) 72oC - 9min R: CCCTTCATTCTTCCTCTATGTTG trnL - F (“c”): CGA AAT CGG TAG ACG CTA CG 95oC - 5min, 94oC - 1min, Taberlet et trnF 58oC - 1min, 72 oC - 2min al., 1991 30s, 28x (94oC - 1min, 52oC - R (“f”): ATT TGA ACT GGT GAC ACG AG 1min, 72 oC - 2min 30s), 72oC - 12min

39

Table 2-2. Microsatellite loci and primer sequences used in this study

Locus Primer sequence (5‟-3‟)* Repeat No. of alleles

Hyg01AT11 F: GGCCTCTTTACCAGTTACGC (AT)11 6 R: GATCGTTTAATTTGTACATAAAGAGTTACG

Hyg02ATT9 F: CCCGAGGCCTATGTTACTCG (ATT) 9 8 R: CCTCCACTCGAACTTGCTGG

Hyg03ATAC7 F: AATTTAGAGCGGAGGGAGGG (ATAC)9 4 R: GTTGATGTACCATCTGATCCCC

Hyg05AT13 F: TAATTTGCTCACCTTCGCCC (AT)13 7 R: CAATAGTTACTTAAATCGCCGATGG

Hyg07AT11 F: AAGTAGTGGCCATGGTTGGG (AT)11 7 R: CCCTCAACCATAAGTCTGGAAGG

Hyg10ATT15 F: AAATAAGAAGAGACCAGACTTTATATCCC (ATT)15 10 R: TTGACAGTTTTGTTTGTGACGC

Hyg19AT11 F: AACAGAAATGAACAGCCGGG (AT)11 8 R: CAACAGAACAAATACAAGCAGCG

Hyg20TC11 F: TGCTCTTTACCCACGTCTGC (TC)11 8 R: TGAATGTTGTAGTTGAGTTGTTTGG

Hyg25ATT14 F: AATTTAGCAAGCTGAGCCCG (ATT)14 12 R: CGTGGATTCTAAGCATGTCAGG

Hyg27ATT15 F: CATAGCTCAAGCCTCAAGCC (ATT)15 13 R: TTCGATTTCTTACTTGCTTCCC

Hyg28AC13 F: TGAATTACACTCCTTGCCGC (AC)13 3 R: CTACGCACAGGAAATTTCAGC

* Note M13 tail (ACG ACG TTG TAA AAC GAC) was added to the 5‟ end of F primer for attachment of florescent dyes

40

Table 2-3. Allelic diversities of microsatellite loci across invasive and native populations of hygrophila

Invasive habitats Native habitats

USA Australia Mexico India Bangladesh S. Florida Texas Carolina

N 8 4 4 8 7 7 9 Hyg01AT11 NA 1 1 1 1 1 6 5 HE 0 0 0 0 0 0.786 0.623 HO 0 0 0 0 0 0.429 0.333 FIS - - - - - 0.455 0.465

Hyg02ATT9 NA 1 1 1 1 1 5 6 HE 0 0 0 0 0 0.735 0.753 HO 0 0 0 0 0 0.286* 0.333 FIS - - - - - 0.611 0.557

Hyg03ATAC7 NA 1 1 1 1 1 2 3 HE 0 0 0 0 0 0.133 0.204 HO 0 0 0 0 0 0.143 0.222 FIS ------0.77 -0.091

Hyg05AT13 NA 2 2 2 2 2 4 6 HE 0.50 0.50 0.50 0.50 0.50 0.633 0.809 HO 1.0* 1.0* 1.0* 1.0* 1.0* 0.286 0.111* FIS -1.0 -1.0 -1.0 -1.0 -1.0 0.569 0.863

Hyg07AT11 NA 2 2 2 2 2 6 6 HE 0.50 0.50 0.50 0.50 0.50 0.806 0.698 HO 1.0* 1.0* 1.0* 1.0* 1.0* 0.286* 0.667 FIS -1.0 -1.0 -1.0 -1.0 -1.0 0.646 0.044

Hyg10ATT15 NA 2 2 2 2 2 6 5 HE 0.50 0.50 0.50 0.50 0.50 0.776 0.525 HO 1.0* 1.0* 1.0* 1.0* 1.0* 0.286* 0.556 FIS -1.0 -1.0 -1.0 -1.0 -1.0 0.632 -0.059

*Deviations from Hardy-Weinberg equilibrium, P<0.05

NA, number of alleles at a locus; HE, expected heterozygosity; HO, observed heterozygosity; FIS, inbreeding coefficient

41

Table 2-3. Continued

Invasive habitats Native habitats

USA Australia Mexico India Bangladesh S. Florida Texas Carolina

N 8 4 4 8 7 7 9 Hyg19AT11 NA 1 1 1 1 2 5 8 HE 0.0 0.0 0.0 0.0 0.245 0.765 0.840 HO 0.0 0.0 0.0 0.0 0.286 0.571 0.667 FIS - - - - -0.167 0.253 0.206

Hyg20TC11 NA 1 1 1 1 1 5 4 HE 0.0 0.0 0.0 0.0 0.0 0.551 0.685 HO 0.0 0.0 0.0 0.0 0.0 0.286* 0.111* FIS - - - - - 0.481 0.838

Hyg25ATT14 NA 2 1 1 1 1 4 10 HE 0.117 0.0 0.0 0.0 0.0 0.684 0.833 HO 0.125* 0.0 0.0 0.0 0.0 0.143* 0.556* FIS -0.067 - - - - 0.791 0.333

Hyg27ATT15 NA 1 1 1 1 1 8 8 HE 0.0 0.0 0.0 0.0 0.0 0.847 0.809 HO 0.0 0.0 0.0 0.0 0.0 0.571 0.889 FIS - - - - - 0.325 -0.099

Hyg28AC362 NA 1 1 1 1 1 3 3 HE 0.0 0.0 0.0 0.0 0.0 0.643 0.494 HO 0.0 0.0 0.0 0.0 0.0 0.286 0.222* FIS - - - - - 0.556 0.550

Mean NA 1.364 1.273 1.273 1.273 1.364 4.909 5.818 HE 0.147 0.136 0.136 0.136 0.159 0.672 0.661 HO 0.284 0.273 0.273 0.273 0.299 0.325 0.424 FIS -0.767 -1.0 -1.0 -1.0 -0.792 0.477 0.328 † AR 1.32 1.27 1.27 1.27 1.35 4.03 4.04 † PA 0.03 0.0 0.0 0.0 0.02 1.73 1.72

*Deviations from Hardy-Weinberg equilibrium, P<0.05 †Calculated using rarefaction method adjusting sample size using 4 individuals (Kalinowski, 2005)

NA, number of alleles at a locus; HE, expected heterozygosity; HO, observed heterozygosity; FIS, inbreeding coefficient; AR, Allelic richness; PA, Private allele richness

42

Figure 2-1. Map showing collection sites of plant samples across invasive and native ranges of hygrophila

43

-500 ln(X|K) ΔK 120

-600 100 -700 80 -800

60 K Δ

ln(X|K) -900 40 -1000

-1100 20

-1200 0 1 2 3 4 5

K Figure 2-2. Inference of the number of unique genetic clusters (K) from Structure simulations for invasive and native populations of hygrophila (K = 1 to 5).

44

Figure 2-3. Assignment of hygrophila individuals collected from invasive and native habitats into district clusters (K = 2), inferred from Structure simulations.

45

-400 12 ln(X|K) ΔK 10 -450

8 -500

6 K Δ

ln(X|K) -550 4

-600 2

-650 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 K

Figure 2-4. Inference of the number of unique genetic clusters (K) from Structure simulations for native populations of hygrophila (K = 1 to15).

46

Figure 2-5. Assignment of hygrophila individuals collected only from native habitats into district clusters (K = 2 to 5), inferred from Structure simulations. Note each color represent a distinct genotypic cluster.

47

CHAPTER 3 EFFECT OF SIMULATED HERBIVORY ON GROWTH OF THE INVASIVE WEED HYGROPHILA POLYSPERMA: EXPERIMENTAL AND PREDICTIVE APPROACHES

The Enemy Release hypothesis (Williams, 1954) predicts that upon arrival in a new habitat, invasive plants experience a loss of associated natural enemies, which plays a significant role in controlling their growth and vigor in native habitat (Keane and

Crawley, 2002). Based on this principle, classical biological control intentionally releases specialized natural enemies in the invaded range to reduce density and spread of an invasive weed (Mueller-Schaerer and Schaffner, 2008). Selection of the most damaging natural enemies is critical to the success of biological control programs. This is important because herbivory-induced compensatory plant growth can nullify the feeding damage of an defoliator (Strauss and Agrawal, 1999; Gadd et al., 2001;

Sun et al., 2009). Therefore, selection of natural enemies should be based on plant responses to herbivory (Raghu et al., 2006b).

Simulation of herbivory through mechanical removal of leaf tissue provides insight about plant compensation and tolerance to actual herbivory (Raghu et al., 2006b). This information can aid in generating an explicit hypothesis about potential damage caused by different levels of herbivory, and can be applied in selecting the most effective guilds of natural enemies (Raghu et al., 2006a). Herbivory simulation provides many advantages over actual herbivory (Baldwin, 1990). For instance, it allows accurate analysis and interpretation of defoliation effects, controls for biotic and abiotic confounding factors (Belsky, 1987), and is less cumbersome, as no insect rearing is involved (Hjalten, 2004). However, concerns have been raised by some scientists that this method fails to initiate induced plant responses that follow actual herbivory

(Capinera and Roltsch, 1980). Despite these shortcomings, Lehtila and Boalt (2004) 48

concluded that herbivory simulation could still be an effective method for approximately imitating „simpler plant responses‟, such as biomass accumulation.

In classical biological control programs, surveys are conducted in native habitats of the target weed to identify promising natural enemies (Harley and Forno, 1992).

Numerous insects and pathogens are generally collected during such surveys. For instance, two mites and 20 insect species were collected as potential biological control agents of the old world climbing fern, Lygodium microphyllum (Cav.) R. Br.

(Lygodiaceae, Pteridophyta) (Goolsby et al., 2003). Several techniques have been proposed for a priori selection and prioritization of promising natural enemies (Harris,

1973; Goeden, 1983; Wapshere, 1985; Harris, 1991; McEvoy and Coombs, 1999;

Forno and Julien, 2000; Davis et al., 2006). An empirical plant growth model is one technique that could aid in understanding the effects of herbivory on growth of an invasive weed, and can be particularly suitable for biological control (Godfray and

Waage, 1991; Kriticos, 2003). For example, one was used by Wilson et al. (2000) to simulate population dynamics of water hyacinth, Eichhornia crassipes (Mart.) Solms,

(Pontederiaceae) and one of its biocontrol agents. Although, empirical models are precise in simulating plant growth, their applicability is limited to a particular ecological situation, as varying growth conditions will alter calculated values of model parameters

(Thornley and Johnson, 1990).

In this study I first conducted a mesocosm experiment to examine the effects of defoliation on growth and biomass accumulation of hygrophila. This was done to determine the critical level of herbivory necessary to achieve a significant reduction in growth of this invasive species. An empirical plant growth model was then developed

49

based on the data obtained during the mesocosm experiment. The goal of this predictive modeling was to evaluate the model‟s usefulness as a tool for a priori assessment of hygrophila‟s response to defoliation.

Materials and Methods

Experimental Set Up

This mesocosm experiment was conducted for eight weeks from April to June

2009 at the Biven‟s Arm Research Center, University of Florida, Gainesville, Florida,

USA. Seven treatments were set up in a randomized complete block design with four outdoor concrete tanks (approx. 900 L capacity) as blocks. The treatments comprised three levels of manual defoliation: 0% (Control), 50%, and 100% with three frequencies of defoliation, once (50%-1x and 100%-1x), twice at four weekly intervals (50%-2x and

100%-2x) or at weekly intervals (50%-8x and 100%-8x). To mimic natural growth conditions, plants were kept completely submerged throughout the experimental duration. Water in those tanks was changed three times per week to prevent algal growth.

Hygrophila plants were collected in April, 2009 from the Santa Fe River at Rum

Island Spring, Columbia County, Florida, USA. Apical shoot cuttings each containing at most 4 nodes (~10 cm) were collected from individual submerged plants. Cuttings were transplanted singly in plastic pots (7 cm x 7cm x 6 cm) containing a mixture of sand and soil (3:2). Pots (n = 315) were assigned randomly to treatments (n = 7, 45 plants/treatment) and transferred immediately to each tank (blocking factor, n = 4).

Plants were initially allowed to grow for two weeks before applying the first treatment. At the start of the experiment, insect defoliation was simulated by cutting all leaves according to the routine mentioned above, either widthwise from the middle or from their

50

bases with sterile scissors to achieve 50% and 100% herbivory, respectively. Assuming that herbivory will only occur on new foliage, repeated damage was simulated by applying designated levels of defoliation to any new growth of plants assigned to a particular treatment at weekly intervals.

Data Collection

For each block (n = 4), five plants were randomly selected per treatment (n = 7) and destructively sampled at the start of the experiment and at weekly intervals to measure the lengths of the tallest stem above ground (cm) and the longest root below ground (cm). Total leaf area (cm2) and average leaf area defoliated per plant per treatment (cm2/plant-treatment) were measured using a LI-3000C Portable Area Meter

(Li-Cor Biosciences, Nebraska, USA). Plants were then dried in an oven at 70oC for 72 hours before measuring stem, root and leaf dry weights (g). Weekly total biomass (g) was calculated as the sum of root, stem and leaf dry weights. Total biomass at the end of the experiment was designated as final biomass. Relative plant growth rate (RGR, average increase in plant dry mass per week) was calculated from weekly change in biomass, using the formula (Eq. 3-1) provided by Hoffmann and Poorter (Hoffmann and

Poorter, 2002).

LnB  LnB RGR  t1 t [Eq. 3-1] (t 1)  t

Where, LnBt+1 and LnBt denote the natural logarithms of total plant biomass at the current (t) and the following (t+1) week.

Model Description

Based on a mass balance approach (Carr et al., 1997), a set of mathematical equations was developed using data collected during the mesocosm experiment. This

51

model was based on the following assumptions: (i) plants grow exponentially with non- limiting space, (ii) optimal conditions for growth existed with no resource limitation, (iii) there was no loss of biomass except through defoliation treatments, and (iv) plant response to defoliation was linear. Four plant growth variables; biomass (B), leaf area

(L), leaf dry weight (BL) and axis dry weight (BA) (axis = stem + root), were calculated using this model. To enable direct comparison, axis dry weight for the mesocosm experiment was calculated by adding root and stem dry weights. Table 3-1 summarizes the variables and parameters used in the model.

new The rate of production of new dry biomass Bt  was calculated as a function of leaf area L and axis dry weight BA at week t (Eq. 3-2). The parameters R and R  t   t  L A were defined as the net leaf and axis photosynthetic rates, respectively and were calculated based on weekly change in leaf area and axis dry weight (Eq. 3-4). For plants subjected to defoliating treatments, net change in new dry mass was calculated by subtracting the amount of leaf dry weight defoliated. Leaf damage can be calculated in terms of leaf area (cm2/week) or leaf weight (g/week) basis. For example, DeLoach and Cardo (1976) measured the adult feeding damage of the water hyacinth biological control agents Neochetina eichhorniae and N. bruchi (Coleoptera: Curculionidae) in terms of leaf area damage per day. In contrast, Quimby et al. (1979) measured the adult feeding damage of the alligator weed (Alternanthera philoxeroides Griseb,

Amaranthaceae) ( hygrophila Selman and Vogt; Coleoptera:

Chrysomelidae) in terms of leaf dry weight damage/day basis. I presented two alternative equations (2A and 2B) to accommodate both forms of damage measurement, either one of which can be used during modeling simulation.

52

In equation 3-2A, LI ,t denotes the amount of leaf area damaged per week

(cm2/week) and is converted to dry mater loss by the Specific Leaf Weight (SLW, g/cm2). SLW is defined as the leaf dry weight BL per unit of leaf area L (Eq. 3-3).  t   t 

dB new t  R .L  R .BA  SLW.L [Eq. 3-2A] dt L t A t I ,t

Alternatively, equation 3-2B captures the damage measurement in dry weight

(g/week) basis. LIW ,t represents the loss of leaf dry weight as a result of defoliation damage, if it occurs on week t.

dB new t  R .L  R .BA  L [Eq. 3-2B] dt L t A t IW ,t

BL SLW  t [Eq. 3-3] Lt

new Bt1,t  LA,t  Lt1   BA  BA    t t1  RL .   RA .  [Eq. 3-4] 7  2   2 

Where, L represents the remaining leaf area after defoliation at week t. A, t

Observations taken at the current and following week were denoted by t and t+1, respectively. This equation was solved simultaneously with values from two treatments,

Control and 100%-8x. The objective of choosing these two treatments was to calculate the contribution to biomass growth from leaf and axis photosynthesis at two extremes of defoliations (following Boote et al., 1983): no leaf damage for control and 100% leaf damage at weekly intervals for 100%-8x. For 100%-8x treatment, the average leaf area

cut while applying weekly defoliating treatments LIt  was subtracted from the total leaf

area Lt  to calculate the leaf area after defoliation LA,t . This enables the model to

53

capture the actual leaf growth during any particular week. For control plants, with no

defoliation, LA,t = Lt .

new As no loss of biomass is assumed, the new growth Bt  is apportioned to develop new leaf or axis. The following two equations (Eqs. 3-5 and 3-6) calculate the net

change in leaf and axis weights, respectively. The partitioning coefficient  L is the

new proportion of weekly new growth Bt  attributed to development of new leaf tissue.

dBL  dB new  t  t    L   [Eq. 3-5] dt  dt 

dBA dB new t  1  t [Eq. 3-6] dt L dt

The partitioning coefficient  L  was calculated using the following equation (Eq.

3-7), as the ratio of weekly change in leaf weight and total biomass. Where, Bt and BLt are total biomass and leaf dry weight at week t, respectively.

BL  BL t1 tt  L  [Eq. 3-7] Bt1  Bt 

 dBt  Finally, change in total biomass   was calculated by adding the changes in  dt  leaf and axis weights (Eq. 3-8)

dB dBL dBA t  t  t [Eq. 3-8] dt dt dt

Model Simulation

The model was simulated on a weekly basis (Δt = 1 week) following the Euler method (Jones and Luten, 1988) (Eq. 3-9).

54

dX t X tt  X t  t [Eq. 3-9] dt

Where value of a plant growth variable, X, at time t+Δt was computed by adding weekly change due to growth and defoliation with its previous value at time t. To enable direct comparison of observed and predicted values, the model was predicted over eight weeks. Herbivory during model simulation also was implemented at weekly intervals. I followed two approaches during simulations. First, the accuracy of model predictions was tested against field observations and second, the model was simulated with actual insect feeding data to test its applicability and usefulness in management decision.

Test of predictive accuracy

To test accuracy of model prediction, for each treatment, a simulation was run with initial values (at t = 0) averaged over all replications. This was done to keep the modeling error (defined as the difference between observation and prediction) independent of the intrinsic variations within replicated observations for each treatment.

Statistical analyses were carried out only on these results. In addition, to calculate the standard error of the prediction, the model also was simulated with randomly selecting

10 plants from each replication and only used during graphical representation of modeling error.

Test of model usefulness

The objective of this explorative approach was to test the model‟s performance and utility in guiding management decisions when simulated with different levels of actual insect feeding data. Feeding damage measured either in leaf area (cm2/week) or leaf mass (g/week) were simulated using equations 2A and 2B, respectively.

Representative feeding rates of four successful biocontrol agents were used for this

55

purpose: N. eichhorniae and N. bruchi (adult feeding rates 86 mm2/day and 75 mm2/day, respectively) (DeLoach and Cordo, 1976), A. hygrophila (adult feeding rate

0.28 mg/day) (Quimby et al., 1979), and the tropical soda apple (Solanum viarum Dunal,

Solanaceae) biological control agent Gratiana boliviana Spaeth (Coleoptera:

Chrysomelidae, larval feeding rate 2.8 mg/day) (R Diaz personal communication).

Assuming constant feeding, all rates were converted to a cm2/week (for leaf area measurements) or g/week (for leaf weight measurements) basis. Initial observations for

10 randomly selected control plants (same plants as used in previous section) were used for this simulation. Relative growth rates (RGR) were calculated post hoc from weekly changes in biomass (Eq. 3-1). The model was simulated for eight weeks to allow direct comparison with field observations.

Statistical Analysis

Mesocosm experiment

All statistical analyses during this study were conducted using SAS version 9.2

(SAS Institute Inc., NC, USA). To determine the effects of defoliation treatments on plant growth, data collected at the end of the experiment were subjected to ANOVA using PROC GLM. Under the randomized complete block design, analyses were conducted with tanks (block) as random effects and herbivory treatments as fixed effects with no interaction terms. The Shapiro-Wilks statistic generated through PROC

UNIVARIATE was used to verify normality of data. Root length data were normalized using a log transformation (log10) and root dry weight, stem dry weight and leaf area were square root transformed. Pairwise comparisons of treatment means were performed using the LS mean procedure of PROC GLM with Bonferroni-corrected t- tests at α = 0.05.

56

Predictive model

Test of predictive accuracy: Model prediction accuracy was evaluated using linear regression and deviation based statistical measures, following techniques reported in

Mayer and Butler (1993), Elbir (2003) and Alagarswamy et al., (2006). Similar to the mesocosm experiment, values obtained only at the end of the simulation were subjected to statistical analysis. The regression analysis was performed using PROC

REG, where variation of the observed and the predicted values were tested against a linear regression line with slope = 1 and intercept = 0 (Analla, 1998). Scatter plots of measured (X) versus calculated (Y) values with a 1:1 line (X = Y) were constructed to visually interpret model agreement with empirical data.

In conjunction with the regression analysis, three commonly used deviation based measures also were used to verify model performance: Root Mean Squared Error

(RMSE) (Eq. 3-12) including its systematic (RMSES, Eq. 3-10) and unsystematic

(RMSEU, Eq. 3-11) components; Modeling Efficiency (EF) (Eq. 3-13) (Loague and

Green, 1991); and Willmott‟s Agreement Index (D) (Eq. 3-14) (Willmott, 1981).

2 0.5  1 N     RMSE   P  O  [Eq. 3-10] S   i i    N i1   

2 0.5  1 N     RMSE   P  P  [Eq. 3-11] U   i i    N i1   

2 2 RMSE  RMSEU  RMSES [Eq. 3-12]

2 N   Oi  Pi  i1   , [Eq. 3-13] EF  1 N 0  EF 1 2 Oi  O i1

57

2 N   Oi  P i  i1   D  1 2 , 0  D 1 [Eq. 3-14] N    P i  O  Oi  O  i1  

Where, O represents the observed value of the ith treatment and P represents i i

 the corresponding value calculated by the model. O is the average of , and Pi is the least squares linear regression of and with slope a and intercept b ( = a +b). N represents the total number of treatments (n = 7).

The limiting values for D and EF are 0 (complete disagreement) and 1 (perfect agreement) (Elbir, 2003; Wallach, 2006). For RMSE, values close to 0 indicate close agreement between observation and prediction (Elbir, 2003). RMSES measures the linear bias of prediction, which is the distance between the linear regression best fit line and the 1:1 line. The unsystematic modeling bias RMSEU, on the other hand, estimates the scatter of individual data points from the linear regression best fit line (Alexandris et al., 2008). A good model is characterized by low RMSEu and RMSES close to total

RMSE (Alexandris et al., 2008). In addition, Student‟s t-tests were conducted using Proc

T of SAS to test statistical difference between observed and predicted means.

The model assumed linearity of plant response across defoliating treatments. To

test this assumption, specific leaf weight (SLW, Eq. 3-3) and partitioning coefficient ( L ,

Eq. 3-7) calculated from randomly selected plants (n = 10 plants/treatment, same plants used to test model‟s predictive accuracy) were subjected to one-way ANOVA, implemented using PROC GLM. Pairwise comparisons of mean SLW and across

58

treatments were implemented using the LS mean procedure with Bonferroni-corrected t- tests (α = 0.05).

Test of model usefulness: Similar to the mesocosm experiment, values obtained at the end of the simulations were subjected to one way ANOVA and pairwise comparison of means following methods mentioned in previous section.

Results

Mesocosm Experiment

Tanks (blocks) had no significant effects on any of the plant growth variables.

Therefore, only the effects of the defoliating treatments are discussed in the following section.

Stem length and stem dry weight

Simulated herbivory affected hygrophila stem length (F6, 130 = 132.43, P < 0.0001).

Increasing levels of herbivory resulted in a reduction in stem length. Control plants were significantly taller than all other plants (Fig. 3-1A). There was no difference in stem length between treatments 50%-1x, 50%-2x, and 100%-1x. Likewise, stem length of plants in 50%-2x, 100%-1x and 100%-2x treatments were not different. Average stem length of plants in 50%-8x was shorter than all other treatments.

Stem dry weight also was affected by manual defoliation (F6, 130 = 38.8 P <

0.0001). However, higher levels of defoliation did not always cause a reduction of stem dry weight. Control plants and 50%-1x treated plants had the greatest stem dry weight

(Fig. 3-1B). Similarly, stem dry weights of 50%-1x plants were not different from those of plants in 50%-2x, 100%-1x and 100%-2x. The same trend also was found for 50%-8x,

100%-2x and 100%-8x.

59

Root length and root dry weight

Overall, defoliation treatments affected hygrophila root length (F6, 130 = 38.3 P <

0.001). However, no differences were observed between most of the treatments, 50%-

2x, 50%-8x, 100%-1x, 100%-2x, and 100%-8x (Fig. 3-1C). Control plants and plants in

50%-1x treatment had the greatest root lengths (Fig. 3-1C). Root lengths of 50%-1x treated plants were not different from those in 50%-2x plants.

Hygrophila root dry weights were affected by defoliation (F6, 130 = 25.1 P < 0.0001).

No differences were observed between control, 50%-1x and 100%-1x treatments (Fig.

3-1D). All other treatments were significantly different from control.

Leaf area and leaf dry weight

The defoliation treatments affected leaf area (F6, 130 = 2472.7, P < 0.0001). Control plants had the greatest leaf area followed by 50%-1x (Fig. 3-1E). Among the completely defoliated plants, the highest leaf area was found in 100%-1x, which was significantly lower than 50%-2x, but higher than 50%-8x. Plants in 100%-8x had the lowest leaf area.

Similar to leaf area, defoliation negatively affected hygrophila leaf dry weight (F6,130

= 804.8, P ≤ 0.0001) (Fig. 3-1F). Control plants had the highest leaf dry weight.

However, no difference in average leaf dry weight was observed between 50%-8x,

100%-1x and 100%-2x. Lowest leaf dry weight was observed in 50%-8x and 100%-8x treated plants.

Final biomass

Final biomass (biomass at the end of the experiment) was affected by herbivory

(F6, 130 = 571.8, P ≤ 0.0001). With no leaf damage, the final biomass of control plants was higher than biomass in any other treatments (Fig. 3-1G). However, there was no difference in the final biomass of plants in 50%-1x and 50%-2x. Subjected to highest

60

level of defoliation, 100%-8x treated plants accumulated less biomass than all other treatments. There was no difference in final biomass between 50%-8x and 100%-2x.

Relative growth rate (RGR)

Effects of defoliation on mean weekly RGR of hygrophila was significant (F6, 130 =

77.9, P ≤ 0.0001). The control plants grew at the highest rate (Fig. 3-1H) with no difference in mean weekly RGR of 50%-1x and 50%-2x treated plants. For plants subjected to 50%-8x, 100%-2x and 100%-8x, treatments, the average RGRs were negative. Lowest RGR was recorded from 100%-2x and 100%-8x treated plants.

Model Simulation Results

Test of prediction accuracy

In general, there was close agreement between observations and model predictions, as indicated by the high r2 recorded for all plant growth variables (Table 3-

2). The slopes of the regressions for the plant response variables were not significantly different than 1 (Table 3-2).

For all treatments, simulated final biomass agreed closely with observed values

(Table 3-2). Although not significantly different, average final biomass was generally under-predicted by the model (Fig. 3-2A). Total RMSE was approximately 15% of the mean biomass observed, indicating that the model captured most of the variation. The

RMSES explained approximately 74% of total RMSE, confirming that this model indeed was accurate in predicting final biomass (Fig. 3-2A). This also was corroborated by the high values of EF and D (Table 3-2).

Similar results were observed for leaf dry weight. Here again, the model in general under-predicted average leaf dry weight (Table 3-2). The low total RMSE

(approximately 22% of mean observed) demonstrated the accuracy of the prediction.

61

The proportion of RMSEU to total RMSE was found to be approximately 84%, indicating high unsystematic error in predicted leaf dry weight (Fig. 3-2B).

For leaf area, there was no difference between observations and predictions. The total RMSE was low and approximately 18% of observed mean, emphasizing prediction accuracy (Table 3-2). Modeling efficiency (EF) and Willmott‟s Agreement Index (D) were high and approaching one, again confirming the accuracy of modeled prediction. On the other hand, high RMSEU, which was approximately 86% of total RMSE, showed intrinsic variation of predictions for individual treatments, from the linear regression best fit line

(Fig. 3-2C).

Predicted axis dry weights were not different from observed values. Total RMSE was low and approximately 19%, indicating that the model was able to capture most of the observed variation (Table 3-2). RMSEU was found to contribute only about 21% of total RMSE, confirming accuracy of prediction. Conversely, a low EF indicated that variation did exist between observation and prediction. The predicted axis weights for control and 100%-1x treatments demonstrated substantial variations from the 1:1 line

(Fig. 3-2D).

Defoliation had a significant effect on specific leaf weight (SLW, F7, 72 = 4.5, p <

0.001). However, except 100%-1x treated plants, no variation in SLW was recorded in

any other treatments (Fig. 3-3A). Similarly, partitioning coefficient  L  also was

significantly affected by defoliation treatments (F7, 72 = 8.7, p < 0.0001). The  L of control, 100%-1x and 100%-2x treated plants were lower than all other treatments.

Lowest was recorded from 100%-1x treated plants (Fig. 3-3B).

62

Test of usefulness

I only presented the results for biomass accumulation and average RGR of hygrophila as these variables are important to assess ability of an insect defoliator to achieve successful control. Different levels of herbivory were found to have significant effects on final biomass of hygrophila (F5, 54 = 8.8, P < 0.0001). No difference between observed and predicted control emphasized the accuracy of model prediction (Fig. 3-

4A). No difference also was seen in biomass accumulation between observed and predicted control and plants defoliated at the A. hygrophila adult feeding rate. Similarly, no difference was found between plants defoliated at either the N. eichhornae or N. bruchi adult feeding rates. Plants defoliated at the G. boliviana larval feeding rate accumulated the lowest biomass (Fig. 3-4A).

A significant effect of herbivory also was observed on weekly average RGR (F5, 54

= 34.9, P < 0.0001). No difference in RGR was found between observed and predicted control plants as well as plants defoliated at the A. hygrophila adult feeding rate (Fig. 3-

4B). Average RGR of plants defoliated at the G. boliviana larval feeding rate was lowest among all treatments (Fig. 3-4B).

Discussion

Mesocosm Experiment

In general, growth and biomass accumulation of hygrophila were negatively affected by defoliation. These results are consistent with similar studies conducted on other invasive plants, for instance, herbivory simulation studies on the invasive liana

Macfadyena unguis-cati L (Bignoneaceae) (Raghu et al., 2006a); Brazilian Peppertree,

Schinus terebinthifolius Raddi (Anacardiaceae) (Treadwell and Cuda, 2007); Pricly

63

acacia, Acacia nilotica ssp. indica Willd. ex Del. (Mimosaceae) (Dhileepan et al., 2009); and Water hyacinth (Soti and Volin, 2010).

As indicated by the defoliation results, hygrophila was generally able to compensate for single or sometimes two defoliation events. For example, there was no difference in stem dry weight between control and 50%-1x (Fig. 3-1A), indicating that plants were able to compensate for a single defoliation of 50%. Similarly, in the case of root dry weight, hygrophila compensated for multiple defoliation events (Fig. 3-1D), as there was no difference between the control, 50%-1x, 50%-2x and 100%-1x. Defoliation damage was possibly buffered by increased leaf (Kovach et al., 1992) and stem photosynthetic activities.

In the case of leaf area, the pattern of plant response observed across treatments showed high correlation with the amount of leaf tissue removed (Fig. 3-1E). However, this relationship did not hold true for leaf dry weight. Here, severity in defoliation did not always result in a proportional reduction in leaf dry weight. For example, although there were differences in leaf area between 50%-8x, 100%-1x and 100%-2x treatments (Fig.

3-1E), corresponding leaf dry weight was not difference (Fig. 3-1F). Variation in SLW observed across treatments can help in explaining this disparity (Fig. 3-3A). Specific leaf weight (SLW), a measure of leaf toughness (Landsberg, 1990; Steinbauer, 2001), is known to play important role in herbivory resistance in plant. For example, Choong

(1996) demonstrated that leaf toughness had significant negative effect on caterpillar feeding damage to Castanopsis fissa Rehder & EHWilson (Fagaceae). In an earlier study, Wright et al. (1989) reported an increase in SLW of water hyacinth leaves

(measured as leaf hardness, g/cm2) following feeding by N. eichhorniae. Similarly, in the

64

case of hygrophila, SLW also tended to increase with increasing defoliation levels, although there was no significant difference across treatments. The short duration (8 weeks) of this experiment may explain the lack of statistical difference in SLW. Based on the above information, I propose the testable hypothesis that defoliation stimulates an increase in SLW in response to increasing levels of defoliation.

Defining the critical level of herbivory necessary to achieve an acceptable reduction of growth and biomass accumulation of hygrophila is crucial for biocontrol success. This weed can grow in dense stands that clog the entire water column, disrupting irrigation, navigation and flood control activities (Schmitz and Nall, 1984;

Sutton, 1995; Woolfe, 1995). Besides, it also is a very competitive plant, capable of rapid response to stress by altering its resource acquisition ability (Botts et al., 1990).

These characteristics have important implications for biological control because a biocontrol agent not only has to reduce total biomass but also should reduce the plant growth rate. Although 50%-1x and 50%-2x treatments affected accumulation of biomass

(Fig. 3-1G), no significant suppression of RGRs were achieved under these levels of defoliation (Fig. 3-1H). Therefore, from a management perspective, an insect defoliator that causes complete defoliation of hygrophila at least at monthly intervals should be selected to achieve biological control success of hygrophila.

Mathematical Modeling

The model was accurate in predicting the observations recorded during the mesocosm experiment (Table 3-2). The prediction for final biomass showed that the model was accurate in capturing the variations observed across treatments. Similar results also were observed for axis dry weight. In contrast, the high proportions of the unsystematic error (RMSEU) for leaf area and leaf dry weight suggested that the model 65

did not perform well for these two variables (Table 3-2). One possible explanation for this variation lies in the way this model was implemented. As mentioned earlier, values

of the model parameters (SLW and L ) were obtained by averaging those calculated for the two extremes of treatments, Control (no leaf damage) and 50%-8x (maximum leaf damage). This was done based on the assumption that average of the two extremes would accurately represent plant responses over all treatments. However, post hoc statistical analysis showed that for SLW and , significant variations existed between average values used and values for individual treatments, indicating non-linearity in hygrophila‟s response to defoliation, which violates the model assumption (Figs. 3-3A and 3-3B). These variations have important implications for model output. For instance, use of a lower value for could have improved prediction of axis dry weights particularly for control and 100%-1x. Similarly, altered values of SLW could also have affected model performance. The previous information suggests that the errors in model prediction were the results of intrinsic alternation of plant behavior rather than flaws in model equations.

As summarized in McClay and Balciunas (2005), in many reported instances, well established biocontrol agents have failed to cause any significant growth reduction of their target weeds. Such failed attempts not only incur substantial loss of resources but also can trigger indirect ecological effects (Powell, 1989; McClay and Balciunas, 2005;

Pearson and Callaway, 2006). Therefore, an a priori evaluation of agent effectiveness can significantly reduce loss of time and resources allocated for extensive host range testing. Model simulation with representative levels of herbivory of four successful biocontrol agents demonstrated that this model can generate valuable insight about

66

agent efficacy (Fig. 3-4). However, it is important to note that the simulation results presented in this study measured the effects of defoliation by a single insect, which is rarely true in nature. This model also provides opportunity to simulate effects of defoliation under different insect densities.

However, because empirical models are very much situation specific, any alteration in growth conditions, such as a change in plant nutrition, can significantly affect model performance. Another limiting factor of this predictive model is that it only can predict the effects of defoliation, thereby seriously limiting its applicability. Plant damage caused by phloem and xylem feeding insects, and non-defoliating chewing insects (e.g. stem borers, leaf miners) would not be represented by the model. In addition, performance of biocontrol agents depends on many complex but important ecological interactions, e.g. inter- or intra- specific competition. Excluding these factors can seriously limit practical application of the model (McFadyen, 1998). However, detailed mechanistic models are not always available, and warrant specialized knowledge to formulate. The simplified approach presented here can be easily used and modified to suit a particular situation. This model, therefore, could provide biocontrol practitioners a useful additional tool for objective selection of the most damaging insect defoliator and also aid in the process of pre-release efficacy assessment of biocontrol agents.

67

Table 3-1. Variables and parameters used in the plant growth model

Notations used in Descriptions Values Units equations

Variables t Time, in week week

Bt Total dry biomass at time t - g/plant new Bt New dry mass growth at time t - g/plant- week 2 Lt Leaf area at time t - cm /plant 2 LA,t Remaining leaf area after defoliation at - cm /plant time t Leaf area removed per week at time t - cm2/plant- LI ,t week Leaf dry weight removed per week at - g/plant- LIW ,t time t week

BLt Leaf dry weight after defoliation at time t - g/plant

BAt Axis dry weight at time t - g/plant Parameters Rate of net weekly leaf photosynthesis / 0.0003a g/cm2-week RL unit area Rate of net weekly stem photosynthesis / 0.2338a g /g-week RA gram SLW Specific leaf weight (leaf weight / unit leaf 0.0020a g/cm2 area) Coefficient of partitioning of new growth 0.6865a g-leaf/ g-  L into leaf weight plant aValues averaged over 100%-8x and Control, two extremes of defoliation treatments.

68

Table 3-2. Results of the statistical analyses used to evaluate model performance

Leaf dry weight 2 Final Biomass (g) Leaf Area (cm ) Axis dry weight (g) Statistical (g) parameters Obs Sim Obs Sim Obs Sim Obs Sim

Mean 0.23 0.21 0.120 0.112 62.510 58.994 0.115 0.102 Max 0.47 0.43 0.286 0.297 133.754 150.432 0.180 0.138 SE (±) 0.053 0.049 0.041 0.04 18.671 20.277 0.013 0.008

Regression Based Measures Slope ± SE 0.88 ± 0.72 0.95 ± 0.12 1.06 ± 0.10 0.56 ± 0.13 (t) (0.94NS) (0.41NS) (0.32 NS) (0.08 NS) Intercept ± SE -0.01 ± 0.02 -0.01 ± 0.02 -7.34 ± 8.004 0.04 ± 0.02 (t) (0.30NS) (-0.36NS) (-0.92NS) (2.43NS) r2 0.97 0.96 0.95 0.78

Deviation based measures Total RMSE 0.0347 0.0264 11.4841 0.0213 RMSES 0.0279 0.0105 4.1633 0.0190 RMSEU 0.0207 0.0242 10.4189 0.0096 EF 0.9382 0.9338 0.9369 0.5730 D 0.9833 0.9830 0.9854 0.8429

Mean test:† p value 0.0797NS 0.3847NS 0.4607NS 0.1313NS

†α = 0.05; Obs, observed; Sim, simulated; NS, not significant; D, Willmott‟s index of agreement; RMSE, root mean square error; RMSEU, unsystematic component of RMSE; RMSES, systematic component of RMSE; EF, modeling efficiency

69

Figure 3-1. Effects (Mean + 1 SE) of defoliation treatments on plant growth parameters of hygrophila. Observations taken at the end of the experiment were subjected to statistical analysis. Bars with same letters adjacent to them are not statistically different as tested by the LS mean procedure of PROC GLM with Bonferroni-t test at α = 0.05.

70

Figure 3-2. Mean observed vs. predicted final biomass (A); leaf dry weight (B); leaf area (C) and axis dry weight (D) summarized across all treatments. Horizontal and vertical bars represent estimated standard errors, respectively for observation and simulation. Solid line show linear regression and dotted line show 1:1 line. Note axis = stem + root. Observed axis weights were calculated by adding respective stem and root dry masses observed during field experiment.

71

Figure 3-3. Specific leaf weight, SLW (A) and partitioning coefficient (B) of plants subjected to different levels of defoliation. Gray bars represent values of treatments and black bar represents value used in the simulation (average of 100%-8x treatment and Control). Bars with same letters adjacent to them are not statistically different as tested by the LS mean procedure of PROC GLM with Bonferroni-t test at α = 0.05. See model description for equations used to calculate SLW (Eq. 3-3) and (Eq. 3-7).

72

Figure 3-4. Weekly progression of observed and predicted biomass (A) and average relative growth rate, RGR (B), when simulated with representative feeding rates of know biocontrol agents. Note for Neochetina eichhornae and N. bruchi feeding rate was calculated in cm2/week basis (simulated using Eq. 3- 2A) and Agasicles hygrophila and Gratiana boliviana, rate of feeding was calculated in g/week basis (simulated using Eq. 3-2B).

73

CHAPTER 4 EXPLORATORY SURVEYS IN NATIVE AND INVASIVE HABITATS TO IDENTIFY NATURAL ENEMIES ASSOCIATED WITH HYGROPHILA

Florida historically has been vulnerable to invasion by exotic and plants

(Ferriter et al., 2006). The aquatic weed hygrophila, Hygrophila polysperma (Roxb.) T.

Anderson (Acanthaceae), is one such plant. It is listed as a federal noxious weed

(USDA, 2006), and a Florida Exotic Pest Plant Council listed Category I invasive species (FLEPPC, 2009). This aquatic plant escaped cultivation by the aquarium trade and is now causing serious problems by displacing native aquatic vegetation (Spencer and Bowes, 1985), primarily in lotic environments (Spencer and Bowes, 1985; Sutton,

1995). Hygrophila is believed to be native broadly to the southeastern Asiatic mainland

(Les and Wunderlin, 1981; Spencer and Bowes, 1985; Schmitz, 1990; Angerstein and

Lemke, 1994).

This herbaceous perennial weed is capable of forming dense stands and can occupy the entire water column, causing disruption in irrigation and flood control systems (Schmitz and Nall, 1984; Sutton, 1995). Laboratory studies revealed that hygrophila can tolerate a wide range of growing conditions and it has been suggested that water temperature in Florida will not pose any restriction on its spread (Spencer and Bowes, 1985). Hygrophila can withstand a wide range of habitat conditions and can grow as a submerged plant, as an emergent or even as a terrestrial plant along river banks (Spencer and Bowes, 1985). Spencer and Bowes (1985) anticipated that hygrophila will successfully compete with native submerged plants in shallow water and eventually may out-compete them in water bodies with a pH ranging between 5.0 and

7.0. In a recent experiment, Doyle et al. (2003) reported that the native Ludwigia repens

J. R. Forst. (Onagraceae) is susceptible to displacement by hygrophila due to

74

hygrophila‟s higher biomass production potential. Hygrophila has a very high growth rate and is capable of rapidly expanding its population ten-fold in a year (Vandiver,

1980). Reproducing vegetatively, rooted small nodes of the easily fragmented stems of hygrophila have the potential to develop new stands. Regrowth potential from these stem fragments even surpasses that of hydrilla (Les and Wunderlin, 1981; Spencer and

Bowes, 1985). Due to its very high reproductive potential and vigor, Les and Wunderlin

(1981) predicted that hygrophila could eventually become a severe problem in Florida waterways. Unfortunately, their prediction has been realized.

Conventional measures do not provide effective control of hygrophila. Mechanical methods may be useful for removing the floating mats, but harvesting increases the number of stem fragments that can be transported to other areas where they can infest new water bodies (Sutton, 1995). Established hygrophila infestations also are difficult to control with registered aquatic herbicides (Grime, 1977; Sutton et al., 1994; Sutton,

1995). Due to presence of cystoliths (calcium carbonate pustules) in its leaves and stems, hygrophila is not a preferred host for triploid grass carp Ctenopharyngodon idella

Val (Cuvier & Valenciennes) (Pisces: Cyprinidae), a polyphagous fish that was introduced into Florida in the 1970s, primarily for controlling hydrilla (Sutton and

Vandiver, 1986; Sutton, 1995).

A noticeable increase in the number of public lakes and rivers with hygrophila has occurred in Florida since 1990 (Langeland and Burks, 1999). One of explanations for the recent hygrophila problem is simply a lack of host specific natural enemies attacking the plant, which would give it a competitive advantage over native flora. Surveys of the

75

natural enemies of hygrophila are needed because there is no information available on potential biological control agents of this aquatic plant (Cuda and Sutton, 2000).

Classical biological control relies on the intentional release of host specific natural enemies collected from the native habitats of the target weed (Freedman et al., 2007).

However, as reported in several studies, adventive or those from the invasive habitat can sometimes provide substantial control of invasive weeds

(Freedman et al., 2007). For example, the immigrant Asian Parapoynx diminutalis

Snellen, reported from Florida and Panama, was found cause occasional damage to hydrilla (Buckingham and Bennett, 1989). Similarly, Cuda et al. (2002) reported that a putatively adventive midge, Cricotopus lebetis Sublette, also was causing widespread damage to hydrilla in Crystal River, Citrus Co., Florida. In another example, Rayamajhi et al. (2008) reported that an adventive scale insect Paratachardina sp. (Hemiptera:

Kerriidae) and a rust fungus (Puccinia psidii G. Wint) affected the above ground biomass allocation of the invasive tree Melaleuca quinquenervia (Cav.) S.T.Blake

(Myrtaceae). Therefore, surveys in Florida were necessary to identify herbivores associated with hygrophila in its invasive habitat. This information may be useful in identifying vacant niches that can be exploited by host specific natural enemies collected from the plant‟s native range.

The objectives of this study were to (i) identify arthropods associated with hygrophila in its invasive habitat;(ii) catalog and georeference historical populations of hygrophila from herbaria records in India; and (iii) identify candidate natural enemies associated with hygrophila in its native range.

76

Materials and Methods

Surveys in Florida

Surveys were conducted in Florida (n = 7 sites, Fig. 4-1) in August 2009 to catalog arthropods associated with hygrophila in its invasive range. Two sampling methods, sweep net and clipped vegetation sampling, were employed during this survey. As hygrophila can grow both as a terrestrial and a submerged plant, both types of plants were collected during each sampling occasion.

Sweep samples (n = 10) were collected using a standard entomological net while walking at a normal pace. For clipped vegetation sampling (n = 10), two PVC quadrates of 0.25 m2 (0.5 m x 0.5 m) each were randomly placed approximately at 5 m intervals along the sampling location. All plants inside the quadrates were cut using hand-held grass clippers. The aerial portion of the hygrophila plants along with the water surrounding each sample was placed in a labeled 1 galon Ziploc® plastic bag and transported to the laboratory. In the laboratory, the plant material in each sample was removed from the plastic bag and placed in a Berlese funnel under a 40w bulb until the leaf material completely dried. The contents of the plastic bags were then poured through a 420 micron brass sieve, preserved in 80% isopropanol and examined under a dissecting microscope. Arthropods collected in these domestic surveys were submitted to the Museum of Entomology, Florida State Collection of Arthropods, Department of

Plant Industry, Florida Department of Agriculture and Consumer Services, for identification.

Catalog and Geoposition of Herbaria Records

In order to identify areas for field surveys, locality information of hygrophila specimens was collected from herbaria in the plant‟s native range. In India, herbaria

77

label information was obtained from The Central National Herbarium, Howrah, India under the Botanical Survey of India (see Appendix A). A heads-up digitizing technique was used to generate geopositions based on the available herbaria label information

(Fig. 4-2). In cases where the herbaria label information was too vague to pinpoint a specific location, the centroid of the polygon described in the herbarium label was chosen. In addition, label information was collected from the Royal Botanic Garden

Herbarium, Kew, United Kingdom (Carol Ellison, personal communication, see

Appendix B).

Exploratory Field Surveys in Hygrophila’s Native Range

Exploratory field surveys were conducted in September 2008 and September 2010 in a range of habitats in India and Bangladesh to collect natural enemies of hygrophila.

In total, 41 sites were surveyed, including 28 sites in the states of West Bengal and

Assam, India and 13 sites in Mymensingh, Bangladesh (Fig. 4-3). The geoposition and altitude of each survey site were recorded. Detailed water analyses (pH, electrical conductivity and dissolved oxygen) were conducted to characterize water quality in the native range. Several collection techniques, e.g., hand-picking, Berlese funnel extraction, sweep and clip vegetation sampling, as well as dissection of plant parts, were used to collect natural enemies. Arthropods collected during surveys were preserved according to standard methods. Specimens were submitted for identification to cooperating systematists with expertise on specific taxa.

78

Results and Discussion

Surveys in Florida

Insects collected from Florida include three beetle species (Coleoptera:

Chrysomelidae, Curculionidae and ), three plant hopper species (Hemiptera:

Cicadellidae) and two moth species (families Crambidae and Noctuidae) (Table 4-1).

All the coleopterans were found in Berlese funnel extractions. No direct feeding damage was observed for any of these species. The weevil, Perigaster cretura Herbst

(Coleoptera: Curculionidae) collected in this study was known to feed on Ludwigia spp.

(Onagraceae) and was considered as a potential biological control agent for water primrose (Freedman et al., 2007). Similarly, the frond-feeding weevil, Stenopelmus rufinasus Gyllenhal (Coleoptera: Curculionidae) collected during this study was released as a biocontrol agent of Azolla spp. (Azollaceae) (Hill, 1998).

The aquatic crambid moth Synclita obliteralis Walker was found to cause substantial damage to submerged hygrophila plants and was collected from all locations surveyed in Florida. As reported in Habeck and Cuda (2010), this insect causes complete defoliation of hygrophila. Like most members of the subfamily , the larva makes a leaf case and feeds from inside the case. However, this insect is polyphagous and is known to feed on nearly 60 plant species (Habeck and Cuda 2010).

Therefore, it cannot be considered as a potential biological control agent of hygrophila.

The hemipteran insects were collected in sweep nets (Table 4-1). The leaf hopper

Draeculacephala inscripta Van Duzee (Hemiptera: Cicadellidae) is known to feed on water lettuce (Pistia stratiotes L., Araceae) and Ludwigia peploides (HBK) Raven

(Onagraceae) (Center et al., 2002).

79

Catalog and Geoposition of Indian Herbaria Records

In total, 64 herbaria records of hygrophila were examined and the locality information/ ecological notes recorded from the Central National Herbarium at Kolkata,

India (Appendix A). The herbarium‟s records indicated that hygrophila was collected from 12 Indian states and the majority of samples (26 of 64, or 41%) were from the state of West Bengal in the eastern part of India (Fig. 4-2). The earliest record dated back to

1910 and at least one sample was collected at an altitude of 1200m.

Label information from 41 specimens was collected from the Kew Herbarium,

Richmond, Surrey, UK (Carol Ellison personal communication, Appendix B). Records were collected from specimens dating back to the 1800s and early 1900s from Pakistan,

Burma, Vietnam, Taiwan, Sri Lanka and Malaysia. This information was helpful in delimiting the native range of hygrophila.

Exploratory Field Surveys in Native Range

In total, 28 sites were surveyed in India, including 15 in the state of West Bengal, and 13 in the state of Assam (Fig. 4-3). Detailed descriptions of surveyed sites are provided in Appendix C. In Bangladesh, surveys were conducted in 13 sites in the region of Mymensingh (see Appendix D for site descriptions).

A number of insects were collected during these surveys in the native habitats of hygrophila. In comparison to its invasive range, substantially higher numbers of herbivores were associated with hygrophila in its native range. Insects collected from hygrophila‟s native range belonged to the Orders: Coleoptera (, Buprestidae,

Carabidae, Chrysomelidae, Coccinellidae, Curculionidae, Dytiscidae, Hydrophilidae,

Noteridae, , and Staphylinidae), Lepidoptera (Crambidae, Noctuidae and

Nymphalidae) and Hemiptera (Cicadellidae, Delphacidae, Meenoplidae and Tingidae).

80

Table 4-2 summarizes the insects collected along with their approximate abundances, trophic status and methods of collection.

Coleoptera

In total, 19 genera of in 11 families were collected during surveys in India and Bangladesh (Table 4-2). The most important species observed to cause direct damage to hygrophila was a leaf mining buprestid beetle Trachys sp. (Fig. 4-4). The genus Trachys is very large (>600 species) and has never been revised (Charles L.

Bellamy personal communication). It is known to occur throughout the entire Palaearctic region east through all of Asia, throughout all of Africa and at least one species occurs in Australia (Bellamy, 2008). This insect was collected from both Assam and West

Bengal in India, as well as Mymensingh area in Bangladesh form terrestrial population of hygrophila. Based on field observations, the buprestid completes its life cycle within a single leaf of hygrophila (Fig. 4-4A). The dorsoventrally flat wedge shaped larva has a characteristic buprestid form with large dorsoventrally flattened head and thoracic regions (Fig. 4-4A). The larva mines entirely inside hygrophila leaves, feeding on leaf tissue from side to side without damaging the upper and lower leaf epidermis, forming a transparent leaf cavity. Larval duration is ~3 weeks. Pupation occurs within the leaf pocket and the pupal duration is ~7 days. The pupa is brownish in color and ~6-7mm long and 2-3 mm broad (Fig. 4-4B). The adult beetle is metallic black in color, 3-4 mm in length and 1-1.5 mm in width (Fig. 4-4C-D). It is important to note that rate of field infestation of this insect was found to be very low. In total, 19 larval specimens were collected during the surveys. As this insect completes its life cycle within a single leaf, the size of the hygrophila leaf may be an important factor limiting its rate of infestation.

The average length and breadth of the leaves found to be infested with this insect

81

ranged between 3 ± 0.5 cm and 1.2 ± 0.2 cm (mean ± SD) , respectively. Infestations were usually observed in the broader and longer lower leaves of the plant.

In an earlier study, Subramaniam (1920) described a congener T. bicolor

Kerremans infesting Butea monosperma (Lam.) Taub (earlier B. frondosa, Fabaceae) in

Mysore, Karnataka, India. The larval and pupal descriptions reported in Subramaniam

(1920) matches closely with those observed in this study. However, the adult characteristics of these two species differ significantly. As described by Subramaniam

(1920), T. bicolor adults were metallic blue with white wavy markings on the elytra. In contrast, the beetle reported in this study is metallic black with no markings on the elytra

(Fig. 4-4D), confirming that this species is not T. bicolor. Although there are no records of Trachys spp. as weed biological control agents, other buprestid beetles were used successfully or tested as potential agents in several weed biocontrol programs. For example, the bronze knapweed root borer, jugoslavica Obenb.

(Coleoptera: Buprestidae) was used as a biological control agent of diffuse knapweed

Centaurea diffusa Lam. (Compositae) (Powell and Myers, 1988). A congener, S. foveola

(Gebler) was tested by Volkovitsh et al. (2009) against Chondrilla juncea L.

(Asteraceae). The buprestid, hyperici Creutzer was released against perforatum L. (Hypericaceae) (Campbell and McCaffrey, 1991). Further studies on

Trachys sp. will be required in order to understand its life cycle and host range. This information will be essential for determining its usefulness as a potential biological control agent of hygrophila.

In addition to the Trachys sp., several other phytophagous leaf beetles were collected during this study (Table 4-2). Available information indicated that congeners of

82

some of these insects have been considered as potential biocontrol agents for other weeds. For example, Scott and Adair (1990) listed Altica sp. as a potential biocontrol agent of the South African weed Chrysanthemoides monilifera (L.) T.Nord.

(Asteraceae). In another case, Kok (2001) reported that the Chrysomelid Cassida rubiginosa Müller successfully controlled the plumeless thistle (Carduus acanthoides L.,

Asteraceae). However, as indicated in Table 4-2, these chrysomelid beetles were collected during sweep sampling or Berlese funnel extraction. Therefore, there is no direct evidence that these insects feed directly and/or exclusively on hygrophila.

A weevil, Bagous luteitarsis Hustache (Curculionidae) also was collected during these surveys (Table 4-2) (O'Brien and Askevold, 1995). This is a semi-terrestrial species in the usually aquatic genus and to date there is no available host plant information (Charles O'Brien, personal communication). The genus Bagous has been important for classical biological control of aquatic weeds (O'Brien and Askevold, 1995).

For example, two species, B. hydrillae O‟Brien and B. affinis Hustache were released as biocontrol agents of hydrilla in the United States. However, B. luteitarsis was only collected once during my surveys in the native range, suggesting that this is not an abundant species. However, considering the importance of this genus in other weed biological control programs and also that this is a semi-terrestrial species, enabling it to attack both terrestrial and submerged forms of hygrophila, further studies are needed to evaluate its biological control potential.

Hemiptera

In total, 10 species in the order Hemiptera, including suborders, Auchenorrhyncha

(families Cicadellidae, Delphacidae and Meenoplidae) and Heteroptera (family

Tingidae) were collected during these surveys (Table 4-2). All the auchenorrhynchan

83

genera were known to be phloem feeders and are major pests of agricultural crops including rice (Misra, 1980; Dash and Viraktamath, 1997; Heinrichs and Barrion, 2004;

Rao and Chalam, 2007; Dai et al., 2009). These insects were collected during sweep sampling and probably use hygrophila as an alternate host. No information on host range is available on the two tingid species (Suborder Heteroptera), Belenus bengalensis Distant and Cantacader quinquecostatus Fieber, collected during this survey (Distant, 1902; Distant, 1909). It is noteworthy that several tingid insects have been used as biocontrol agents or reported as feeding on invasive weeds in several classical weed biocontrol programs (summarized in Julien and Griffiths, 1998). For example, Teleonemia scrupulosa Stal (Hemiptera: Tingidae) was released in a number of countries to control Lantana camara L. (Verbenaceae). The above information suggests that tinged insects could be successful biocontrol agents. Therefore, further studies will be necessary to confirm the feeding damage of B. bengalensis and C. quinquecostatus on hygrophila.

Lepidoptera

Three lepidopteran species, an aquatic leaf cutter Parapoynx bilinealis Snellen

(Crambidae: Acentropinae) (Fig. 4-5), and two foliage feeders, Nodaria sp. (Noctuidae:

Herminiinae) (Fig. 4-6) and Precis almana L (Nymphalidae) (Fig. 4-7) were collected during the native range surveys. All three species were observed to cause direct feeding damage to hygrophila.

Like its congener, P. diminutalis Snellen and another Acertopine moth S. obliteralis

Walker, larvae of this insect make cases with hygrophila leaves and feed internally. The fully grown larva is cream colored, ~15 mm in length and with conspicuous branched tracheal gills on all body segments except the prothoracic region (Fig. 4-5A). Presence

84

of branched gills along the body segments is the diagnostic character for this genus

(Habeck, 1974). Duration of the larval stage was ~3 weeks. The larva feeds by scraping leaf tissue from the inner surface of the case, rendering it transparent. However, in some cases, they were observed to stretch out form the larval case to feed on nearby leaves. No damage to the stem, however, was observed. Pupation occurs in the leaf case, and the leaf case with the pupa floats on the water surface. The duration of the pupal stage was ~7 days. The adult moth (wing span ~10 mm, Fig. 4-5B) is yellowish brown in color with conspicuous white wavy strips on the wings. The insect caused substantial damage to the submerged hygrophila plants. Available records on host range suggest that insects of this genus tend not to be host specific (for example,

Habeck, 1974). The laboratory host range study of the adventive moth P. diminutalis demonstrated that although the moth clearly preferred hydrilla, it fed and developed on other aquatic plants as well (Buckingham and Bennett, 1989). In addition, as summarized in Habeck (1996) other Asian Parapoynx species, for example, P. stagnalis

Zeller, P. fluctuosalis Zeller and P. crisonalis Walker also have wide host ranges. Based on this information, it seems unlikely that P. bilinealis is specific to hygrophila. However, as reported in Buckingham and Bennett (1989), hydrilla was the preferred host of P. diminutalis under multiple choice settings, conducting similar tests with P. bilinealis may generate information regarding its preference for hygrophila.

A noctuid moth, Nodaria sp. also was recorded during these surveys and was found to cause substantial damage to emergent and terrestrial populations of hygrophila

(Fig. 4-6). This moth was recorded across all locations surveyed in the native range and was observed to cause complete defoliation of hygrophila plants (Fig. 4-6C). The genus

85

Nodaria is widely distributed throughout Asia (Hampson, 1894). The fully grown semilooper larva is ~35-40 mm long and ~2-3 mm wide with a reddish brown dorsal surface and green ventral surface (Fig. 4-6B). The dorsal surface of the larva changes color from green to reddish-brown with successive molts. Based on field and laboratory observations, this insect has five instars and total duration of the larval stadium is ~14 days. The larva consumes the entire leaf, leaving only the midrib intact and can cause complete defoliation of the plant (Fig. 4-6C). Multiple larvae have been observed feeding on the same plant. Pupation occurs on the plant within a pupal case made with

3-4 hygrophila leaves tied together by silk. The pupa is brownish black in color, ~10 mm in length and 5 mm in width. Duration of the pupal stage is ~7 days. The adult moth is grayish-brown with indistinct blackish antemedial, postmedial and subterminal wavy lines on the wings; the wingspan ~25mm. The presence of ochreous spots on the subterminal lines suggests a similarity to N. externalis Guenée, a congener and also the type specimen of the genus (Hampson, 1894) (Fig. 4-6D). However, the lack of any thickened middle antennal segments, a diagnostic character for the identification of N. externalis (Bailey, 2007), confirmed that the species collected in this study is not N. externalis (Fig 4-6D). Field observations confirmed that this insect can be very damaging (Fig 4-6C), and is probably specific to hygrophila as no feeding was observed in any nearby plants. Further studies are needed to confirm its host range and determine its biological control potential.

In addition, a nymphalid butterfly, Precis almana L (= Junonia almana), commonly known as the Peacock Pansy butterfly, was found feeding on hygrophila (Fig 4-7). This insect, which was collected from both India and Bangladesh, caused complete

86

defoliation (Fig. 4-7C). However, because this is a polyphagous species, known to feed on wide range of plants (Kehimkar, 2008), it has no value for biological control of hygrophila.

Pathogens

In addition to the aforementioned insects, a very damaging aecial rust fungus

(Pucciniales: Pucciniaceae) that completely kills hygrophila plants (hereafter hygrophila rust) was recorded during surveys in India and Bangladesh (Fig. 4-6). Rust infections were observed at all locations surveyed, suggesting that it could be an important natural enemy of hygrophila.

Historically, rust fungi have been used effectively in classical weed biological control programs (summarized in Charudattan, 2001). Release of Puccinia chondrillina

Bubak and Syd. to control the invasive skeleton weed Chondrilla juncea L. (Asteraceae) in Australia was the first record of successful weed biological control with a rust fungus

(Cullen et al., 1973). Recently, Ellison et al. (2008) reported Puccinia spegazzinii de

Toni as a biological control agent of Mikania micrantha H.B.K. (Asteraceae). Rust fungi also have been used successfully as weed biocontrol agents in the US. For example, P. curduorum Jacky was reported to provide effective control of musk thistle, Carduus thoermeri Weinmann (Asteraceae) (Baudoin et al., 1993). In another example, the rust fungus Puccinia jaceae var. solstitialis Savile was used as a biological control agent of yellow starthistle (Centaurea solstitialis L., Asteraceae).

Most rust fungi have a complex life cycle, involving both sexual (aecial and pycnial spores) and asexual (uredenia, telial and basidial spores) stages (see Kolmer et al.,

2009 for detailed description of a rust life cycle). A rust life cycle involving all five spores is known as macrocyclic. Rust fungi, where the sexual and asexual stages occur in two

87

alternate hosts are classified as heteroecious rusts. This type of life cycle is observed in the wheat stem rust, Puccinia graminis f. sp. tritici Pers (Roelfs, 1985). A prerequisite for a rust fungus to be a potential biological control agent is that it must be autoecious, where the rust completes its life cycle within a single host. For example, P. jaceae var. solstitialis is macrocyclic autoecious rust. However, the rust life cycle also can be microcyclic, involving three or fewer spore stages. For example, P. spegazzinii is a microcyclic, autoecious rust with only telio and basidiospores (Evans and Ellison,

2005).

In case of the hygrophila rust, only the aecial (Fig 4-6A-B) and pycnial (Fig. 4-6C) stages were observed in the field. Repeated attempts to inoculate healthy hygrophila plants with aeciospores from infected plants failed to initiate infection. However, laboratory observations confirmed that aeciospores germinated on tap-water agar medium (personal observation). This information suggests that the hygrophila rust may involve multiple hosts (or have a heteroecious life cycle).

In an earlier study, Thirumalachar and Narasimhan (1954) reported that the aecial stage of the heteroecious leaf rust, P. cacao McAlp. occur on Hygrophila spinosa

(Acanthaceae), a congener of H. polysperma. The common grass Hemarthria compressa (L.f.) R.Br. (Poaceae) is the primary host of P. cacao. They reported that on

H. spinosa, the rust develops a systemic infection and the infected shoots are paler in color. Interestingly, similar observations also were made on hygrophila. Laundon (1963) provided descriptions of all rust fungi infecting of Hygrophila spp., at that time. When compared, the measurements of aecial and pycnial spores, collected from H. phlomoides, H. salicifolia and H. spinosa (data compiled by Carol Ellison, CABI, UK) as

88

reported in Laundon (1963) matches closely with those of the hygrophila rust (Table 4-

3). These results suggested the hygrophila rust also could be the sexual stage of P. cacao. Having a wide aecial host range is not uncommon for heteroecious rusts (Kolmer et al., 2009). However, cross inoculation studies to test the ability of aeciospores from hygrophila to initiate infection in Hemarthria compressa is necessary to ascertain the identification of this rust fungus.

Conclusion

Substantially fewer arthropods were associated with this plant in its invasive range, which is consistent with the „Enemy Release Hypothesis‟ (Williams 1954). In

Florida, hygrophila experiences very little herbivore damage and no host specific insects were collected. A number of insects, including two caterpillars (P. almana and Nodaria sp.) that defoliate emerged plants, an aquatic caterpillar (P. bilinealis) feeding on submerged hygrophila, and a leaf mining beetle (Trachys sp.) were collected during surveys in India and Bangladesh, the native range of hygrophila. Some of these insects, in particular, P. bilinealis, Nodaria sp. and Trachys sp. hold promise as potential biocontrol agents of hygrophila. Further studies are necessary to determine their host ranges and specificity to hygrophila.

In addition, several other insects, including the weevil B. luteitarsis and two chrysomelid beetles, Altica sp. and Cassida sp. also were collected during native habitats of hygrophila. As these insects were collected during Berlese funnel extraction, no direct feeding damages were observed for these species. However, considering that congeners of these insects have been investigated as biological control agents, further studies are justified to ascertain their potential as classical biological control agents of hygrophila.

89

A very damaging aecial rust fungus (Puccinia sp.) also was collected during these surveys. Although initial studies suggested that this rust could be the aecial stage of the heteroecious P. cacao, detailed cross-inoculation studies, involving its primary host H. compressa, are necessary to confirm its identity,

90

Table 4-1. Insects collected from hygrophila during surveys in Florida

Approx. Trophic Methods of 1 2 3 References Abundance level collection

Coleoptera Chrysomelidae Paria sp R Leaf feeder Berlese Balsbaugh, 1970 Curculionidae Perigaster cretura Herbst U Leaf feeder Berlese Freedman et al., 2007 Stenopelmus rufinasus U Leaf feeder Berlese Hill, 1998 Gyllenhal Elmidae vittata Melsheimer U Berlese Hillsenhoff, 1973 Hemiptera Cicadellidae Hortensia similis Walker C Sap feeder Sweep Genung and Mead, 1969 Draeculacephala inscripta C Sap feeder Sweep Center et al., 2002 Draeculacephala robinsoni R Sap feeder Sweep Genung and Mead, 1969 Lepidoptera Crambidae Synclita obliteralis Walker C Leaf feeder Hand coll. Habeck and Cuda, 2010 Noctuidae Spodoptera frugiperda Smith U Leaf feeder Hand coll. Capinera, 1999

1Approximate Abundance: R = Rare-collected 3 times or less; U = Uncommon-collected 4-10 times; and C = Common-collected regularly 2Trophic level information for herbivorous species does not imply that species included on this list were actually observed using hygrophila as a food source 3Methods of collection: Sweep = Sweep net sampling; Berlese = Berlese funnel extraction and Hand coll. = larva collected from field and reared to adult

91

Table 4-2. Insects collected from hygrophila during surveys in India and Bangladesh, native habitats of hygrophila

Approx. Trophic Methods of Taxonomy 1 2 3 References Abundance level collection

Coleoptera Anthicidae Anthelephila mutillaria R Scavenger? Sweep Kejval, 2003 Saunders Buprestidae Trachys sp. Fabricius U Leaf miner Hand coll. Bellamy, 2008 Carabidae Bembidion sp R Predator Sweep Clivina sp. Latreille U Predator Sweep Tachys sp R Predator Sweep Chrysomelidae Altica sp. U Leaf feeder Sweep Scott and Adair, 1990 Aspidomorpha sp. R Leaf feeder Sweep Nakamura and Abbas, 1987 Cassida sp. A U Leaf feeder Sweep Friedli and Bacher, 2001; Kok, 2001; Cassida sp. B U Leaf / root Sweep Ghate et al., 2003 feeder Chaetocnema sp. R Leaf feeder Sweep Wapshere, 1990 Lema sp. A R Leaf feeder Sweep Kumari and Lyla, Lema sp. B R Leaf feeder Berlese 2001 Pachnephorus sp. U Leaf feeder Sweep Shah et al., 1990 Philipona sp. U Leaf feeder Sweep Veenakumari et al., 1997 Coccinellidae Harmonia sp. C Predator Sweep Curculionidae Bagous luteitarsis Hustache R ? Berlese O'Brien and Askevold, 1995 Hydrophilidae Helochares sp. U Predator Berlese Merritt and Paracymus sp. U Predator Berlese Cummins, 1996 Regimbartia sp. U Predator Berlese Scarabaeidae Rhyssemus sp. Mulsant R Scavenger? Sweep Staphylinidae Philonthus sp. Stephens R Predator Sweep

1Approximate Abundance: R = Rare-collected 3 times or less; U = Uncommon-collected 4-10 times; and C = Common-collected regularly 2Trophic level information for herbivorous species does not imply that species included on this list were actually observed using hygrophila as a food source 3Methods of collection: Sweep = Sweep net sampling; Berlese = Berlese funnel extraction and Hand coll. = larva collected from field and reared to adult

92

Table 4-2. Continued

Approx. Trophic Methods of Taxonomy 1 2 3 References Abundance level collection

Hemiptera Cicadellidae Nephotettix sp. C Sap feeder Sweep Misra, 1980 Cofana spectra Distant C Sap feeder Sweep Cofana unimaculata Signoret C Sap feeder Sweep Hecalus sp. U Sap feeder Sweep Dash and Viraktamath, 1997 Scaphomonus indicus Distant U Sap feeder Sweep Dai et al., 2009 Delphacidae Nilaparvata sp. C Sap feeder Sweep Misra, 1980 Rao and Chalam, Perkinsiella sp. Kirkaldy C Sap feeder Sweep 2007 Tingidae Belenus bengalensis Distant R Sap feeder Sweep Distant, 1909 Cantacader quinquecostatus R Sap feeder Sweep Distant, 1902 Fieber Lepidoptera Crambidae Parapoynx bilinealis Snellen R Leaf cutter Hand coll. Chen et al., 2006; Mathew, 2006 Noctuidae Nodaria sp. C Leaf feeder Hand coll. Hampson, 1894 Nymphalidae Precis almana L C Leaf feeder Hand coll. Kehimkar, 2008

1Approximate Abundance: R = Rare-collected 3 times or less; U = Uncommon-collected 4-10 times; and C = Common-collected regularly 2Trophic level information for herbivorous species does not imply that species included on this list were actually observed using hygrophila as a food source 3Methods of collection: Sweep = Sweep net sampling; Berlese = Berlese funnel extraction and Hand coll. = larva collected from field and reared to adult

93

Table 4-3. Comparison of aecial rust collected from Hygrophila polysperma with that collected from H. phlomoides, H. salicifolia and H. spinosa by Laundon, (1963)

Aecia (diam.) Aeciospores (diam.) Pycnia (diam) Pycniospores

Puccinia sp. collected from 228.14 - 398.06 µm 18.43 – 28.41 µm 103.41 - 120.05 µm Paraphysate hygrophila

Laundon 200 - 400 µm 15 - 30 µm 80 - 100 µm Paraphysate (1963)*

*Data complied by Carol Ellison

94

Figure 4-1. Survey sites in Florida (n = 7) to document natural enemies associated with hygrophila in its invasive habitat

95

Figure 4-2. Distribution of hygrophila in India based on herbaria records (n = 64) collected from the Central National Herbarium, Kolkata, India. Heads-up digitizing technique was used to generate geoposition of herbaria label information.

96

Figure 4-3. Survey sites in India (n = 28) and Bangladesh (n = 13). In India, surveys were conducted in the states of West Bengal (n = 15) and Assam (n = 13).

97

Figure 4-4. Trachys sp. (Coleoptera: Buprestidae) collected from H. polysperma; A. Larva mining inside the leaf cavity, inset: close-up larva; B. Pupa inside the leaf cavity; C-D. Ventral and dorsal views of the adult beetle. Photo credit A. Mukherjee

98

Figure 4-5. Larva (A) and adult (B) of Parapoynx bilinealis (Lepidoptera: Crambidae) collected from H. polysperma. Note the presence of branched tracheal gill on larval body – a genus characteristic. Photo credit A. Mukherjee

99

Figure 4-6. Nodaria sp (Lepidoptera: Noctuidae) collected from H. polysperma; A. Larva feeding on hygrophila leaves; B. Pupa; C. Feeding damage; D. Adult moth. Photo credit A. Mukherjee

100

Figure 4-7. Precis almana (Lepidoptera: Nymphalidae) collected from H. polysperma. A. Larva; B. Pupa; C. Feeding damage; D. Adult butterfly. Photo credit A. Mukherjee

101

Figure 4-8. Rust fungus (Puccinia sp.) collected from H. polysperma. A. rust infected hygrophila plant; B. Cross section of aecia; C. Cross section of pycnia. Photo credit A. Mukherjee

102

CHAPTER 5 PHYTOPARASITIC NEMATODES ASSOCIATED WITH THE RHIZOSPHERE OF HYGROPHILA

As part of the survey to collect natural enemies of hygrophila, phytoparasitic nematodes present in the rhizosphere of hygrophila were extracted, counted and identified. To compare the diversity of phytoparasitic nematodes in invasive habitats, similar surveys were conducted in Florida. The specific objective of this study was to assess the diversity of the phytoparasitic nematode fauna associated with the rhizosphere of hygrophila in its native and exotic ranges.

Materials and Methods

Sampling and Enumeration of Nematodes

In September 2008, exploratory field surveys (n = 19) were undertaken in a range of locations in the states of West Bengal (n = 12, sites Ind-1 to Ind-12) and Assam (n =

7, sites Ind-13 to Ind-19) in India (Fig. 5-1). Except for two sites in West Bengal (Ind-2 and Ind-8), all samples were collected from natural areas. For sites Ind-2 and Ind-8 samples were collected from irrigation channels in agricultural fields. Each survey site was geopositioned and assigned a unique accession number. Two soil cores containing roots of hygrophila, each of 10 cm3 (10 cm diameter x 10 cm height), were collected at

~10 m intervals from each survey site. All samples were collected from shoreline ~2 m from the edge of the water. Cores (n = 2) collected from each survey site were pooled before extraction of nematodes. Nematodes were extracted following the technique of

Handoo and Ellington (2005), and identification of phytoparasitic nematodes to genus, and in some cases species, were performed at the Plant Health Diagnostic Laboratory,

Department of Agricultural Entomology, Bidhan Chandra Krishi Viswavidyalaya, West

Bengal, India.

103

In August 2009, similar surveys were conducted in Florida (n = 7) to characterize the diversity of phytoparasitic nematodes associated with hygrophila in its exotic range

(Fig. 5-2). Samples were collected from natural areas in five counties, including Alachua

(site Fl-7), Broward (site Fl-1), Dixie (sites Fl-5 and 6), Osceola (sites Fl-3 and 4), and

Pinellas (site Fl-2). Using a metal trowel, two soil cores (~10 cm3) containing hygrophila roots were collected from each survey site at ~10 m intervals. Similar to native habitats, samples were collected from shoreline to ~2 m from the edge of the water. Soil cores (n

= 2) collected from each site were thoroughly mixed before extraction of nematodes.

The geoposition of each survey site was recorded. Nematodes were extracted using aforementioned methods and identified to the genus level in the Nematode Assay laboratory, Entomology and Nematology Department, University of Florida.

Assessment of Nematode Diversity

The diversity of the phytoparasitic nematofauna was assessed for each sampling site (n = 19 for native habitat, n = 7 for exotic habitat). In addition, data from all sampling sites within the native and invasive ranges were pooled to calculate the overall diversity of the nematodes. Following techniques reported in Bernard and Schmitt (2005),

Shannon diversity ( H', Eq. 5-1) and evenness ( EH , Eq. 5-2) indices were calculated to measure the α diversity (within site diversity) of each sampling site and habitat

(Magurran, 2004) using the following equations.

H'  p (ln p )  i i [Eq. 5-1]

EH  H'/ ln S [Eq. 5-2]

104

Where, pi is the relative abundance of each species, calculated as the proportion

th of individuals of the i species ( ni ) to the total number of individuals ( N ) in the

n community, p  i . S denotes the total number of species present in the community or i N

the species richness. The range of values for EH is 0 and 1, with 1 being complete evenness.

For each range, the results of diversity analyses and nematode densities (number of nematodes/100 cm3 soil) were graphed. In addition, the Kruskal-Wallis analysis of variance (hereafter Kruskal-Wallis test, Corder and Foreman, 2009), was used to test the difference in density and diversity of phytoparasitic nematodes between exotic and native ranges. Statistical tests were performed using the open source statistical software R (version 2.11.1, http://cran.r-project.org) at α = 0.05.

Using the Morisita-Horn index of community similarity ( CMH , Eq. 5-3), cluster analysis of nematode assemblages was performed on all sampling sites within each range following the unweighted pair group average (UPGMA) method (Bernard and

Schmitt, 2005). CMH is a measure of β diversity (between site diversity), which calculates the similarity in species composition between two sites. Waldo (1981) investigated a number of similarity indices and recommended the use of because it is not influenced by the effects of sample size and species diversity. The limiting values of are 0 (completely dissimilar) and 1 (completely similar). The Morisita-

Horn index of community similarity is calculated by the following equation.

2 a .b  C   i i MH d  d  N  N  a b   a b  [Eq. 5-3]

105

th Where, ai and bi are the i species of sites A and B, respectively. N a and N b

represents the number of individuals collected, respectively from site A and B. da (and

) is calculated as follows d  a 2 / N 2 d b a  i a .

Results

Nematode Diversity

Native range: In total, eight phytoparasitic nematode species, representing seven genera, were collected from India (Fig. 5-3A). The number of nematode species at sampling sites varied between two (sites Ind-7, 8, 11, and 15) and seven (site Ind-12).

Densities of nematodes extracted (number /100 cm3 soil) varied between 94 (site Ind-2) and 1130 (site Ind-12) (Fig. 5-1B), with an average (mean ± SD) of 609.3 ± 293.8 nematodes/100 cm3 soil (Fig. 5-3B, black bar). The Shannon diversity ( H') of nematodes in the native range (pooled data) was 1.8, with sampling sites ranging

between 0.3 (site Ind-8) and 1.5 (site Ind-6) (Fig. 5-3C). Overall, a high evenness ( EH =

0.9, black bar Fig. 5-3D) of nematode distribution was found across native habitats. The

value calculated among sampling sites ranged between 0.4 (site Ind-8) and 1.0 (site

Ind-11).

Exotic range: In total, 10 phytoparasitic nematode genera were collected from Florida

(Fig. 5-4A), with seven genera collected from site Fl-7 and one genus from site Fl-2.

Nematode densities were found to be low in most of the sites, with an average of 141.9

± 307.7 nematodes/100 cm3 soil with the highest density (830 nematodes/100 cm3 soil) found at site Fl-6, located in the Dixie County, Florida (Fig. 5-2 and 5-4A). The highest

Shannon diversity index was calculated from site Fl-7 ( = 1.3), with an overall invasive range of 0.8 (Fig. 5-4C). Evenness ( ) of nematode distribution across

106

exotic habitats was low (0.4, Fig. 5-4D black bar). Because only a single genus was

collected from site Fl-2 (Fig. 5-4A), both calculated H' and EH were zero (Figs. 5.4C and 5.4D).

There was no difference in between exotic and native ranges (Kruskal-Wallis

2 test,  =3.531, p = 0.06). Similarly, no difference was observed in ( = 0.242, p =

0.62). In contrast, densities of phytoparasitic nematodes in the native range were significantly higher than in the exotic habitat ( = 8.863, p = 0.003).

Phytoparasitic Nematodes Recorded

For both native and exotic ranges, the proportions of individual taxa (genus or species level for India, genus level for Florida) collected from the rhizosphere of hygrophila are shown in Tables 5-1 and 5-2, respectively.

Native range: In total, eight species of phytoparasitic nematodes, all belonging to the order Tylenchida were collected from India (Table 5-1). Among all the taxa, the rice root knot nematode, Meloidogyne graminicola Golden & Birchfield, was the dominant species collected across all sampling sites (n = 19), representing ~34% of all phytoparasitic nematodes (Table 5-1). The spiral nematode Helicotylenchus sp., represented ~17% of all nematodes collected. Another root knot nematode, M. incognita

Chitwood was collected from eight sampling sites (Ind - 3, 5, 6, 13-15, 17, and 19) and represented ~13% of all phytoparasitic nematodes recovered. Other species collected were Criconemoides sp. (~6%), Hirschmanniella oryzae Van Breda de Hann (~11%),

Hoplolaimus indicus Sher (~1%), Rotylenchulus reniformis Linford & Oliveira (~9%) and

Tylenchorhynchus mashhoodi Siddiqi and Basir (~9%) (Table 5-1).

107

Based on the Morisita-Horn index of similarity, sampling sites across India can be divided into two groups with low similarity in phytoparasitic nematofauna (< 0.2, Fig. 5-

5). Among all sampling sites, highest similarity was observed between Ind-7 and Ind-15

(similarity index of 0.9, Fig. 5-5, data not shown). In both cases, two species of nematodes, Helicotylenchus sp. and M. graminicola were collected with approximately equal densities. Overall, similarity indices across native sampling sites found wide variation in the nematode fauna associated with hygrophila roots.

Exotic range: Genus level nematode taxa, representing two orders (Triplonchida and

Tylenchida), were collected from Florida (Table 5-2). Across all the sampling sites (n =

7), the order Tylenchida, with eight genera, was found to be the most diverse. Among

Tylenchida, the genus Helicotylenchus was collected from three sampling sites (Fl-5, 6, and 7) and constituted ~76% of all phytoparasitic nematodes collected. In addition, the genus Tylenchorhynchus constituted ~16% of all phytoparasitic nematodes collected.

All other phytoparasitic nematode genera were represented at lower proportions (Table

5-2). The other order, Triplonchida was represented by two genera, Trichodorus and

Paratrichodorus, collected only from site FL-1.

The Morisita-Horn index of similarity among sampling sites across Florida was found to be generally low (Fig. 5-6). Maximum similarity (~0.9, Fig. 5-6, data not shown) was recorded between sites FL-6 and FL-7. A somewhat lower similarity index (0.9) was found between sites FL-2 and FL-5. In contrast, site FL-1 was distinctly different than all other sites in Florida with no similarity in phytoparasitic nematodes collected.

Discussion

Assessment of nematode diversity across ranges demonstrated that significantly higher densities of phytoparasitic nematodes are associated with roots of this weed in

108

its native habitat. For both ranges, a number of phytoparasitic nematodes collected from the rhizosphere of hygrophila are considered as pests of important agricultural and horticultural crops (Tables 5-1 and 5-2). For instance, the rice blind root knot nematode

(Hirschmanniella oryzae Van Breda de Hann), lance nematode (Hoplolaimus indicus

Sher) and stunt nematode (Tylenchorhynchus mashhoodi Siddiqi and Basir), collected in the native range of hygrophila are considered as major pests of rice (Oryza sativa L.)

(CABI, 2005). In the invasive range, species of the sheath nematode (Hemicycliophora de Man), lance nematode (Hoplolaimus Sher), as well as stunt nematode

(Tylenchorhynchus Cobb) are known as important crop pests (Anderson et al., 1991,

CABI, 2005, Fortuner and Nickle, 1991). This is important because earlier studies demonstrated that invasive weeds can act as alternate hosts for important crop pests, including fungal pathogens (Wisler and Norris, 2005), insects (Seal, 2004) and nematodes (Davis et al., 2006). Important from the perspective of this study, Davis et al., (2006) demonstrated that the invasive weed tropical spiderwort (Commelina benghalensis L., Commelinaceae) can act as an alternative host for the peanut root knot nematode, M. arenaria (Neal) Chitwood. The high densities of nematodes found in the root zone of hygrophila, particularly in its native range, suggest that this weed potentially could act as an alternative host of these important plant parasitic nematodes.

Further studies involving inoculation with phytoparasitic nematodes to assess performance of hygrophila against susceptible host plants can provide further insight about its suitability as a transitional or alternative host. Overall, this study, demonstrated for the first time the root association of plant pest nematodes with the invasive weed hygrophila.

109

Table 5-1. Phytoparasitic nematode species recorded from the root zone of hygrophila in the India

Order, Proportions of individuals Taxon (Sites collected) Z

Tylenchida Criconemoides sp. Taylor 0.06 (4,5,7-11, 14,15) Helicotylenchus sp. Steiner 0.17 (1-11, 13-19) Hirschmanniella oryzae Luc &Goodey 0.11 (3,5,6,8,9,11,12, 19) Hoplolaimus indicus Sher 0.01 (1,4,5,19)

Meloidogyne graminicola Golden & 0.34 (1,3,5-7,12,14,15,18) Birchfield

Meloidogyne incognita Chitwood 0.13 (3, 5,6,13-15,17,19) Rotylenchulus reniformis (juvenile) 0.09 (2,4-6,8, 10,12,15, Linford & Oliveira 16) Tylenchorhynchus mashhoodi Siddiqi 0.09 (1,8,9,13) and Basir

ZNumbers represent corresponding sites indicated in Fig. 5-1

110

Table 5-2. Phytoparasitic nematode genera recorded from the root zone of hygrophila in the Florida

Order Proportions of individuals Taxon (Sites collected)Z

Triplonchida Trichodorus Cobb 0.001 (1) Paratrichodorus Siddiqi 0.002 (1)

Tylenchida Helicotylenchus Steiner 0.763 (5,6,7) Hemicriconemoides Chitwood & 0.003 (4) Birchfield Hemicycliophora de Man 0.041 (3,6,7) Hoplolaimus Daday 0.001 (7) Meloidogyne Goeldi 0.019 (7) Mesocriconema Andrassy 0.007 (2,5,7) Pratylenchus Filipjev 0.002 (2) Tylenchorhynchus Cobb 0.161 (3,6,7)

ZNumbers represent corresponding sites indicated in Fig. 5-2

111

Figure 5-1. Survey sites in India (n = 19). Each site was assigned a unique accession number. Symbols are graduated based on average no of nematodes collected per 100cm3 of soil.

112

Figure 5-2. Survey sites in Florida (n=7). Each site was assigned a unique accession number. Symbols were graduated, with bigger size and darker color symbols represent higher nematode densities (number /100cm3 of soil)

113

Figure 5-3. Number of species, density, Shannon diversity (H‟) and evenness (EH) calculated across sampling sites in India. Black bars represent values for native habitat (India, pooled data). Number labels on x-axis correspond to site numbers in Fig. 5-1. N = native habitat. *Note that for Fig. B, black bar denotes the average number of nematodes/100cm3 soil (609.3 ± 293.8), calculated across all sampling sites.

114

Figure 5-4. Number of genus, density, Shannon diversity (H‟) and evenness (EH) calculated across sampling sites in Florida. Black bars represent values for exotic habitat (Florida, pooled data). Number labels on x-axis correspond to site numbers in Fig. 5-2. E = Exotic habitat. *Note that for Fig. B, black bar denotes average number of nematodes/100cm3 soil (141.9 ± 307.7), calculated across all sampling sites.

115

Figure 5-5. Cluster analysis of phytoparasitic nematode assemblage based on Morisita- Horn index of community similarity from sampled sites in India (see Fig. 5-1 for site locations)

116

Figure 5-6. Cluster analysis of phytoparasitic nematode assemblage based on Morisita- Horn index of community similarity from sampled sites in Florida (see Fig. 5-2 for site locations)

117

CHAPTER 6 PRIORITIZING AREAS IN THE NATIVE RANGE OF HYGROPHILA FOR SURVEYS TO COLLECT BIOLOGICAL CONTROL AGENTS 1

Background

The „Enemy Release Hypothesis‟ (Williams, 1954) predicts that upon arrival in a new geographic area, an invasive plant species experiences a decrease in regulation by natural enemies, which leads to an increase in abundance compared to the native range (Keane and Crawley, 2002). Based on this hypothesis, classical biological control of invasive weeds attempts to reunite target weeds with their specific top-down regulators, thereby controlling them in an environmentally rational and cost effective manner (Mueller-Schaerer and Schaffner, 2008). Biological control of invasive weeds has a long history, as indicated in Julien and Griffiths (1998) and documented in more recent literature (Cuda et al., 2008; Forno et al., 2000; Myers and Bazely, 2003; Page and Lacey, 2006). Contrary to earlier thoughts about ineffectiveness of aquatic insects to manage submerged weeds (Jolivet, 1998; Lodge, 1991; Wilson, 1964), a number of studies have documented instances of successful use of aquatic insects in biological control (Balciunas and Burrows, 1996; Buckingham and Bennett, 1996; Creed and

Sheldon, 1993; Cuda et al., 2002; Grodowitz et al., 2004; Wheeler and Center, 2007). In fact, a recent analysis suggests that biological control of aquatic weeds has a higher success rate than biological control of terrestrial weeds (Cuda et al., 2008).

Foreign exploration is an important step in a classical biological control program

(Harley and Forno, 1992), where surveys are conducted in the native range of an invasive weed to identify potential biological control agents. As discussed by Julien and

1 Reprinted with permission from Mukherjee et al. 2011

118

White (1997), climate matching between exotic and native habitats is an important consideration before conducting such surveys because climatic incompatibility can severely limit the establishment and effectiveness of biological control agents

(Robertson et al., 2008). A number of studies have shown that in suboptimal climatic conditions, biological control agents either failed to establish or did not develop vigorous populations (Byrne et al., 2002; Dennill and Gordon, 1990; Hoelmer and Kirk, 2005;

McClay and Hughes, 1995; Stewart et al., 1996; Wapshere, 1983; 1993). The use of climate matching techniques provides an effective way to prioritize suitable areas in the native range to conduct surveys for natural enemies (Byrne et al., 2002; Dhileepan et al., 2006; Fiaboe et al., 2006; Goolsby et al., 2003; Robertson et al., 2008; Senaratne et al., 2006).

The use of ecological niche models to predict species geographic distributions has received increased attention in recent years (Fiaboe et al., 2006; Peterson and Shaw,

2003; Robertson et al., 2008; Williams et al., 2009). The Maximum Entropy Species

Distribution Model (MaxEnt), developed by Phillips (2006), is one such algorithm that attempts to predict a species‟ ecological niche using point (latitude, longitude) locations and environmental layers. Niche based modeling approaches characterize environmental parameters associated with known occurrences of a species. Based on that information, the models predict ecological space suitable for a species‟ long term survival (Phillips et al., 2006). In addition to MaxEnt, several other models have been developed for presence-only ecological niche modeling (Reviewed in Elith and

Leathwick, 2009). Compared to other models, e.g. GARP (Genetic Algorithm for Rule- set prediction) (Stockwell and Peters, 1999), MaxEnt offers more flexibility because it

119

generates continuous rather than binary output. In addition, recent studies have shown that MaxEnt consistently outperformed other distribution models (Elith et al., 2006;

Phillips et al., 2006), but see McNyset and Blackburn (2006) and Peterson et al. (2007) for alternative views.

Practical solutions for long term control for hygrophila are not available (Cassani,

1996; Schmitz and Nall, 1984; Sutton, 1995; Vandiver, 1980). On the other hand, considering several biological and economic attributes of this weed, Cuda and Sutton

(2000) reported that biological control may be a viable option for managing hygrophila.

In order to collect natural enemies of hygrophila, exploratory field surveys were conducted in India and Bangladesh in 2008 - 09 (see Chapter 4). However, the lack of specific information about the native distribution of hygrophila makes it difficult to prioritize areas for future survey efforts.

Therefore, to identify areas of the native range that are climatically similar to the exotic North America range of this weed, MaxEnt was used to develop predictive distribution maps in India and Bangladesh based on point occurrences from the southern United States and Mexico. The accuracy of predicted native distribution was examined using occurrence records collected during native range surveys.

Materials and Methods

Species Occurrence Data

Two independent occurrence datasets were used in this study. The first set, the

Exotic Range Data (ERD), was used to generate all prediction models (Fig. 6-1). The

ERD included 115 spatially unique point occurrences (97 occurrences from the United

States and 18 occurrences from Mexico). Points were considered spatially unique if they did not fall within in the same grid cell (cell size = 2.5 acrminutes, ~5 km2 at the

120

equator) (Joyner et al., 2010). Occurrences in the United States included 92 points from

Florida (from 25 counties), two points from Texas, one point from Mobile County,

Alabama and 1 point from Berkeley county South Carolina. Most US data points came from my field collections. Data also were retrieved from herbaria label information from the Florida Museum of Natural History‟s herbarium library, University of Florida, the

Institute of Systematic Botany‟s herbarium, University of South Florida

(http://florida.plantatlas.usf.edu), as well as from the Nonindigenous Aquatic Species

Database of the United States Geological Survey

(http://nas.er.usgs.gov/queries/default.aspx). Google EarthTM (http://earth.google.com/) was used to obtain coordinate locations for herbaria specimens lacking latitude/longitude data by selecting a point based on the described collection site. In those cases, where label descriptions were too vague to pinpoint a location, a point in the center of the described polygon was chosen.

All Mexican occurrences of hygrophila (n = 24) were sourced from a recent publication by Mora-Olivo et al. (2008). Although, Kasselman (1994) earlier reported presence of hygrophila in Mexico, the report by Mora-Olivo et al. (2008) provides the first record of its Mexican distribution. Distribution records, collected from herbaria specimens, showed the spread of hygrophila throughout the Tamaulipas region of

Mexico (Mora-Olivo et al., 2008).

One of the major assumptions of niche modeling is that the occurrence localities are drawn from source habitats, where the species can maintain its population without immigration (Phillips et al., 2006). Given the wide geographic distribution of hygrophila

121

in its exotic range, it was assumed that, the used occurrence localities are truly representing the realized niche of this weed.

The second set of occurrence data, Native Range Data (NRD), was comprised of point locations (n = 55) collected during my surveys in India (Fig. 6-3) and Bangladesh

(Fig. 6-4). In total, 42 occurrences were from India, including 24 points from West

Bengal and 18 points from Assam. The remaining 13 points were from the Mymensingh region of Bangladesh (Fig. 6-4). NRD were never part of the any model building process as the objective of my study was to prioritize areas to explore for natural enemies in native range based solely on the climate in its exotic range. Therefore, NRD was used only to objectively validate the native prediction.

Environmental Data Layers

Altitude and 19 bioclimatic (bioclim) variables available freely from the

WORLDCLIM database (http://www.worldclim.org) were used in this study. The ESRI

2.5 arc-minute resolution climate layers were generated by interpolating very high resolution (30 arc second or ~1 km2) average climate data collected from weather stations around the world (Hijmans et al., 2005). The Bioclim variables included in this study are typically used in ecological niche modeling studies (Williams et al., 2009).

Derived from monthly values of temperature and precipitation, Bioclim variables represent annual trends. In this study, no water specific predictor variables were chosen. This is because, in addition to its aquatic habit, hygrophila often grows as an emergent or terrestrial plant in moist habitats (Spencer and Bowes, 1985). Besides, information on water specific variables, like flow rate, pH, dissolved oxygen, etc. are not easily available. During my surveys in the native habitat, emergent or terrestrial growth of hygrophila was observed predominantly. In addition, earlier studies, e.g. McNyset

122

(2005) and DeVaney et al. (2009) have shown that environmental layers that are not water specific were successful in predicting distribution of aquatic species.

Ecological Niche Modeling

To enable statistical verification of the predictive ability of the algorithm, 10 random partitions (named as Partition- 1 to Partition- 10, hereafter referred to as crossvalidation partitions) of the ERD were made with 70% of the points (n = 115) as training data to generate models and the remaining 30% (n = 49) as independent test data for extrinsic verification of predictive accuracy. Crossvalidation is a common technique followed in many niche modeling studies (Anderson and Martinez-Meyer, 2004; Peterson et al.,

2003; Peterson and Shaw, 2003; Phillips et al., 2006). MaxEnt version 3.3.2 (Phillips et al., 2006) was used in this study. For each training set, MaxEnt was run with 10 replications with 30% of points as random test percentage (for internal model testing) and the average prediction was used for statistical analysis. In the MaxEnt settings, the random seeded sub-sampling procedure was chosen as the replicated run type. This ensured that before each replicated run, separate sets of training and intrinsic test points were randomly sampled from the input dataset without replacement. All other parameters were kept at their default settings. Two prediction maps, one each for exotic and native regions, were generated per training dataset.

MaxEnt requires that all climate layers be in the same spatial extent. Therefore, two mask layers (with „0‟ values) were used to restrict the backgrounds of model predictions respectively into Exotic (covering continental US and Mexico; Envelope:

43.04N/122.96W top left, 14.54N/69.92W bottom right) and Native (covering India and

Bangladesh; Envelope: 37.04N/68.17E top left, 6.83N/114.21E bottom right) regions.

The exotic mask also delimited the area for generation of pseudo-absence data during

123

the model building process. Pseudo-absences are localities, chosen randomly from the geographical area on which the model was built, used to provide information about available environmental conditions across the area of interest (Phillips et al., 2009;

Raes and Steege, 2007).

In addition, MaxEnt also was run with the full set of occurrence localities of the

ERD, giving it all available information ensuring better prediction. This was done for final visualization of the predicted distribution of hygrophila. The default logistic output format in which predictive values were scaled from 0 (Not suitable) to 1 (Highly suitable) was chosen.

In MaxEnt‟s prediction, every pixel in the study area receives a nonnegative probability with higher values representing a higher environmental suitability (Phillips et al., 2006). Therefore, the minimum predictive value received by any known native occurrence point (hereafter lowest presence threshold per Pearson et al., 2007) was used as the cutoff to eliminate pixels unsuitable for growth of hygrophila. Pixels with predictive values lower than this threshold were treated as absent. Values above this threshold were classified into 5 subcategories based on thresholds calculated in terms of % omission of known native occurrences. The subcategories were (i) Low (<10% omission), (ii) Mid Low (25% omission), (iii) Mid High (50% omission), (iv) High (75% omission) and (v) Very High (> 75% omission). Maps were color coded with darker colors representing higher environmental suitability.

Statistical Validation of Model Accuracy

To verify that MaxEnt predictions were significantly better than random, a threshold dependent one-tailed binomial test was followed based on omission rate

(proportion of test points falling in pixels not predicted suitable) and Fractional Predictive

124

Area (FPA, fraction of all pixels predicted present) (Anderson et al., 2002; Phillips et al.,

2006). The null hypothesis of the binomial test states that the model is no better than one selected at random from all sets of models with equal FPA (Phillips et al., 2006).

MaxEnt generates continuous predictions. However, conversion of continuous predictions to binary is required for implementing the binomial test. The lowest training presence threshold, smallest non-zero predictive value received by any training locality, was used for converting continuous logistic output of MaxEnt into presence/absence

(Phillips et al., 2006). It is important to note that the lowest training presence threshold was calculated based on a particular training partition and was different from the lowest presence threshold mentioned in previous section. After implementing, predictive values lower than this threshold was treated as absent and the rest as present. As NRD was not part of the model building process, for each data partition, the same threshold value calculated from the ERD training points was used for both exotic and native distribution maps. The histogram tool of DIVA-GIS 5.4 software (http://www.diva-gis.org) (Hijmans et al., 2001) was used to calculate the FPAs.

The proportion of test points (ntest) falling outside the predicted area (outtest) was used as the extrinsic omission rate (outtest/ntest) as per Anderson et al. (2003). For the exotic prediction, each test dataset (30% randomly chosen test data) was used to verify accuracy of the prediction generated by its corresponding training dataset. However, instead of using separate test datasets, full NRD was used to verify all native projections as it was independent of data used for model building. Proc Freq of SAS®

9.0 (SAS 2002) was used to conduct the binomial test.

125

In addition, to the binomial test, predictive performance of the model was evaluated using the threshold independent receiver operating characteristic (ROC) plot following techniques reported in Philips et al (2006). ROC analysis has been popularly used for evaluation of a variety of modeling approaches (Beaumont et al., 2009; Elith et al., 2006), but see Lobo et al. (2008) for potential problems associated with ROC analysis. The area under the curve (AUC) score of ROC plot ranges from 0 to 1, with

0.5 indicating that a model is not better than random and 1 indicating 100% predictive accuracy. Implementing ROC analysis for presence-only data require pseudo-absences

(background data) as random instances (Phillips et al., 2006). A sufficiently large number of background data points are necessary to truly represent all sets of available environmental conditions (Raes and Steege, 2007). However, when implemented with pseudo-absences, AUC no longer achieves a perfect score of 1, but 1-a/2, where a is the fraction of the geographic area (generally unknown) covered by the true distribution of the species (Phillips et al., 2006).

For my purpose, two background data sets were first generated, each consisting of 10,000 points, selected randomly from the native and exotic masks, respectively. For each habitat, AUC was calculated from the independent test localities and pertinent background data set using the R 2.11.1 statistical software (www.r-project.org) following methods described in MaxEnt‟s tutorial (available from www.cs.princeton.edu/~schapire/maxent/). For exotic predictions, 10 independent test data sets were used for AUC analysis for models generated by complementary train data sets. On the contrary, the entire NRD, being independent of the model building process, was used to calculate AUC of all models. Based on the AUC value, model

126

performance was evaluated following classification developed by Thuiller et al. (2006) and also used in Beaumont et al (2009) , where AUC < 0.8 is poor model; 0.8

Results and Discussion

Statistical Tests Results

The threshold dependent binomial test results for exotic habitats were highly significant for all partitions of ERD (Table 6-1; z = 5.145 – 5.831, P < 0.0001, one tailed). FPAs varied from 0.1059 (Partition - 3) to 0.0633 (Partition - 1) with 0.0856 as the average. For all partitions (n = 10), omission rates were consistently low indicating prediction accuracy (average 0.018 ± 0.021, Table 6-1). Similarly, AUC values calculated on independent test points for all partitions were very good (0.95

(Table 6-1). The average AUC calculated was high (0.991 ± 0.003), demonstrating that for all partitions of ERD, MaxEnt truly predicted independent test points as present.

The binomial tests for native predictions also were highly significant (P< .0001) for models generated by all ERD data partitions (Table 6-2), indicating that MaxEnt was highly accurate in predicting the known native distribution of hygrophila (Table 6-2, average omission rate = 0.02 ± 0.006). Overall, the average FPA observed was low

(0.367 ± 0.129), showing that only a small portion of India and Bangladesh was predicted suitable. All partitions consistently showed low omission rates (Table 6-2).

Likewise, AUCs calculated for all partitions, also were good (0.9

0.936 ± 0.013.

The frequency histogram constructed with the probabilities received by NRD (n =

55) also showed that predicted native distribution truly identified areas with known

127

occurrences of hygrophila (Fig. 6-2). As evident, approximately 65% (n = 36) of the

NRD received a suitability score > 0.5. On the other hand, only approximately 9% (n =

5) of known localities have received <0.1 suitability (Fig. 6-2).

The relative importance of environmental variables, based on % contribution in creating distribution model of hygrophila was found to vary across data partitions. On an average, bio18 (Precipitation of warmest quarter) contributed most (42.16% ± 1.02) followed by Alt (Elevation, 19.90% ± 0.69), bio 4 (Temperature seasonality 15.76% ±

0.73) and bio15 (Precipitation seasonality, 14.25 ± 0.38). However it is important to note that, considering the high correlation among environmental variables, the contributions of individual variables should be taken with caution.

From these results, it can be concluded that MaxEnt predictions for all data partitions were accurate and significantly better than random. As mentioned earlier, binomial tests involve threshold dependent conversion of continuous output into binary prediction. The subjectivity in choosing a threshold, however, adds uncertainty to this test. Although there are no general rules for threshold selection, Phillips et al. (2006) suggested that the choice should be based on one the following; (i) number of training points used, (ii) predictive value received by any training point, or (iii) on modeling perspective. Following this suggestion, the minimum predictive value received by any training locality was used as the threshold. The same threshold calculated for a particular data partition was replicated for both native and exotic predictions.

Native Range Predictions

The native range predictions (results of full data set run) obtained for India and

Bangladesh are discussed separately. The lowest presence threshold was used to exclude pixels unsuitable for growth of hygrophila. To prioritize future survey areas,

128

prediction maps were color coded based on % omission of known native occurrences

(Figs. 6-3 and 6-4).

India: The northeastern part of India was predicted to be highly suitable for the growth of hygrophila (Fig. 6-3). Two states in India, Assam and West Bengal, were found to be particularly favorable for growth of this plant. Presence was predicted throughout Assam on both sides of the Brahmaputra River with mid high to very high suitability (Fig. 6-3).

The northern part of West Bengal, which adjoins Assam and Bihar, received very high predictive values. On the contrary, low suitability was reported in regions just south of the Ganges River in West Bengal. Other Eastern Indian states, where distribution of hygrophila was predicted, included parts of Meghalaya, , some parts of Manipur and Nagaland. Other Indian states where this plant was predicted to occur included the northern part of Bihar, Jharkhand, and parts of Uttar Pradesh. On the eastern coast of

India, low suitability was predicted in Orissa along the course of the Mahanadi River.

Similar results also were observed in Punjab and Himachal Pradesh border. Here, the presence of hygrophila was predicted with lower probability along the Sutlej River.

Bangladesh: Most of Bangladesh was predicted to be suitable for growth of hygrophila

(Fig. 6-4). The areas in the northern region, including parts of Dinajpur, Rangpur,

Jamalpur, and Mymensingh were predicted to be highly favorable. In particular,

Northern and Eastern parts of Sylhet were predicted to be very highly suitable (Fig. 6-4).

High suitability of occurrence also was predicted throughout Kishoreganj. High probability values were predicted in the northern and western areas of Bogra. Similar results were reported for northern and central parts of Rajshahi, Dhaka, Tangail and

129

Comilla. Areas further south, Barisal, parts of Noakhlai also were predicted to be highly suitable.

In this “transferral” application of niche modeling (Phillips and Dudik, 2008),

MaxEnt successfully predicted the native distribution of hygrophila based on occurrences from the invaded range. Concerns have been raised by scientists regarding uncertainties as a result of the tendency of ecological niche models to overfit training data while projecting species distribution in novel habitats (Pearson et al., 2006;

Phillips and Dudik, 2008). As highlighted earlier by Peterson et al (2007) and also in

Phillips and Dudik (2008), MaxEnt is prone to such data overfitting. Considering this issue, the results obtained during this study need to be interpreted with caution.

Although, high accuracy of prediction was recorded for known native occurrences, it could again be an artifact of model over fitting.

In this study, invasive range occurrences were used to predict the native distribution. This approach is novel and is less frequently applied in niche modeling studies. However, one of the major problems associated with this approach is that to be successful, niche models assume that species occurrences are drawn from source habitats which may not be true for all invasive species. A source habitat is where a species can self sustain and is in equilibrium with the existing environment. On the contrary, an invasive species may not be in equilibrium with the existing environment in invaded habitats and may require repeated „immigrational subsidy‟ (Grinnell, 1917). In that case, occurrences drawn from invasive habitats will severally limit predictability of a niche model. As mentioned earlier, it was assumed that invasive occurrences are true representatives of realized niche of this species. This assumption was based on

130

hygrophila records over a wide geographic range, suggesting that it is very well adapted to climatic conditions and is self sustaining in its invaded range (Cuda and Sutton,

2000). However, this assumption may not hold true for certain range shifting invasive species (Elith et al., 2010; Trethowan et al., 2010), which could limit applicability of this technique.

As mentioned earlier areas for future surveys were prioritized based on thresholds calcualted in terms of % omission of known native occurrences collected during my surveys. The primary advantage of this method is that it provides an objective basis for threshold selection. To my knowledge, this is the first time this method has been applied in the context of biological control. However, the major disadvantage is that the choice of threshold is solely based on my known knowledge of distribution, which may represent only a small fraction of true native distribution of hygrophila. In that case, the known occurrences based threshold will limit suitable native habitats that should be searched for biocontrol agents.

Conclusion

MaxEnt was accurate in predicting the known native distribution of hygrophila.

Hygrophila was predicted to occur in India (Fig. 6-3) mostly in the northeastern regions of the country. In previous trips, surveys for natural enemies were conducted in Assam and West Bengal, covering most of the highly favorable areas. Based on the final distribution map (Fig. 6-3), new areas were prioritized for conducting future surveys.

These include areas in Meghalaya and Tripura along the Bangladesh border, areas along the Ganges River in Northern Bihar and Utter Pradesh, and locations in Orissa across the Mahanadi River. Most parts of Bangladesh were found to be particularly suitable for growth of this plant, especially areas in the northern plains (Fig. 6-4). During

131

my previous trip to Bangladesh, small area in Mymensingh area was surveyed, leaving room for further and more extensive surveys as indicated in the final distribution map

(Fig. 6-4). Prioritized areas for future surveys include locations in Dinajpur, Rangpur,

Jamalpur, Sylhet, as well as northern parts of Bogra and Rajsahi.

MaxEnt has been used in various aspects of species distribution modeling and has been shown to have predictive accuracy across broad geographic and taxonomic scales

(Baldwin, 2009; Bartel and Sexton, 2009; Boubli and Lima, 2009; Buermann et al.,

2008; Tinoco et al., 2009; Yost et al., 2008). In this study, this modeling technique was applied for the first time as a tool to identify suitable native areas for foreign exploration of natural enemies of hygrophila. As pointed out by Robertson et al. (2008) and reiterated in other studies (Byrne et al., 2002; Hoelmer and Kirk, 2005; Manrique et al.,

2008), climate matching is an important tool for selecting biocontrol agents. The high accuracy of MaxEnt‟s predictions of known native distribution reported here suggests that this modeling approach can be applied to the field of classical biological control.

132

Table 6-1. Result of threshold dependent binomial tests of omission and AUC for predictions with invasive occurrences

Binomial test Data Ext. Test AUC partitions Omm. Rate FPA Omm. Points Z (outtest/ntest) (outtest) (ntest)

Partition- 1 0.063 2 34 0.059 5.145** 0.987† Partition- 2 0.084 1 34 0.029 5.488** 0.990† Partition- 3 0.106 0 34 0.000 5.831** 0.995† Partition- 4 0.084 1 34 0.029 5.488** 0.990† Partition- 5 0.085 0 34 0.000 5.831** 0.992† Partition- 6 0.088 0 34 0.000 5.831** 0.994† Partition- 7 0.084 1 34 0.029 5.488** 0.987† Partition- 8 0.078 1 34 0.029 5.488** 0.989† Partition- 9 0.092 0 34 0.000 5.831** 0.992† Partition- 10 0.091 0 34 0.000 5.831** 0.995†

Average 0.0856 0.018 0.991†

** P< 0.0001 †Very Good

133

Table 6-2. Result of threshold dependent binomial tests of omission and AUC for predictions with native occurrences

Binomial test Data Ext. Test AUC partitions Omm. Rate FPA Omm. Points Z (outtest/ntest) (outtest) (ntest) Partition- 1 0.567 0 55 0.000 7.416** 0.948† Partition- 2 0.304 0 55 0.000 7.416** 0.926† Partition- 3 0.263 0 55 0.000 7.416** 0.954†† Partition- 4 0.522 0 55 0.000 7.416** 0.929† Partition- 5 0.522 0 55 0.000 7.416** 0.928† Partition- 6 0.404 0 55 0.000 7.416** 0.927† Partition- 7 0.202 1 55 0.018 7.147** 0.947† Partition- 8 0.274 0 55 0.000 7.416** 0.944† Partition- 9 0.339 0 55 0.000 7.416** 0.912† Partition- 10 0.270 0 55 0.000 7.416** 0.945†

Average 0.367 0.002 0.936†

** P< 0.0001 ††Very Good †Good

134

Figure 6-1. Occurrences of hygrophila in the US and Mexico (Exotic Range Data or ERD, n = 164). ERD is comprised of point occurrences from the US (n = 140) and Mexico (n = 24).

135

Figure 6-2. Frequency histogram of probabilities received by known native occurrences of hygrophila.

136

Figure 6-3. Projected distribution of hygrophila in India. Warmer colors show high environmental suitability. Northeastern part of India was predicted to be highly suitable.

137

Figure 6-4. Projected distribution of hygrophila in Bangladesh. Warmer colors represent higher suitability. Northern region was found to be particularly suitable for growth of this weed.

138

CHAPTER 7 SUMMARY

Hygrophila, Hygrophila polysperma (Roxb.) T. Anderson (Acanthaceae), is an aquatic invasive plant in the southern United States. It is capable of forming dense stands and can occupy the entire water column, causing disruption in irrigation and flood control systems. This weed can tolerate a wide range of environmental conditions and can grow as a submerged plant, as an emergent or even as a terrestrial plant along river banks. Laboratory and mesocosm experiments demonstrated that it can survive in a wide range of growing conditions, and it has been suggested that hygrophila will successfully compete with native submerged plants and may eventually out-compete them in shallow water bodies with a pH ranging between 5.0 and 7.0.

Currently available control measures are costly and do not provide effective control of this invasive weed. Considering several biological and economic attributes of this aquatic weed, earlier studies have reported that biological control can be a viable option for managing hygrophila. However, no information about the natural enemies of this aquatic plant was available until recently.

In this study, the prospects for biological control of this invasive weed were examined. The primarily objectives were to investigate the genetic variability within the invasive and native populations of hygrophila, study the effects of artificial defoliation on plant growth, conduct surveys in the native range to collect and identify the plant‟s natural enemies, study the diversity of phytoparasitic nematodes in the rhizosphere of hygrophila, and develop niche based prediction models to prioritize areas in the native range for future surveys.

139

Genetic variability of invasive and native populations of hygrophila was analyzed using a combination of chloroplast DNA sequences (psbM-trnD spacer, rpl16 gene, trnS-trnG spacer and trnL-trnF spacer) and microsatellite DNA markers (n = 11). This study demonstrated that no phylogenetically informative variation exists in chloroplast

DNA regions across the invasive and native populations of hygrophila. Structure simulations with microsatellite data from both invasive and native ranges clearly demonstrated that the invasive populations of hygrophila are genotypically identical, suggesting that the invasive population may have originated form a single source followed by subsequent secondary invasions. Considering the history of invasion of hygrophila, this result provided initial evidence that US populations could be the source for introduction to Australia and Mexico. However, detailed studies involving more samples per population are necessary to improve the resolution of the results obtained during this study.

Herbivory simulation studies, through mechanical removal of leaf tissue, can provide valuable insight about plant compensation and tolerance to actual defoliation. A mesocosm experiment was conducted to examine the effects of defoliation on growth and biomass accumulation of hygrophila and to determine the critical level of herbivory necessary to achieve a significant reduction in growth of this invasive plant. Based on the data collected during the experiment, an empirical plant growth model was developed to examine the usefulness of a model based approach for a priori examination of plant response to defoliation. The mesocosm experiment showed that defoliation significantly influenced growth and biomass accumulation of hygrophila. The empirical plant growth model also accurately simulated plant growth response to

140

herbivory across treatments. Based on the results of the mesocosm experiment, an insect defoliator that causes complete defoliation of hygrophila at least at monthly intervals should achieve successful reduction of biomass and growth rate of hygrophila.

Ability of the simple mathematical model to predict the effects of defoliation on hygrophila suggests that it could be a useful tool in selecting the most effective biocontrol agent.

As no information about natural enemies of hygrophila was available, several locations were surveyed in the invasive (Florida, n = 7) and native (India and

Bangladesh, n = 41) ranges of hygrophila. These surveys demonstrated that substantially higher numbers of arthropods were associated with this plant in its native range. In Florida, hygrophila experiences very little herbivore damage and no host specific insects were found. In contrast, several insect natural enemies, including two caterpillars (P. almana and Nodaria sp., Lepidoptera) that defoliate emerged plants, an aquatic caterpillar (P. bilinealis, Crambidae) that attacks submerged hygrophila leaves, and a leaf mining beetle (Trachys sp., Buprestidae) were collected during surveys in the native range in India and Bangladesh. Some of these insects, in particular, P. bilinealis,

Nodaria sp. and Trachys sp. are potential candidates for biocontrol of hygrophila.

Further studies are needed to examine their host ranges and impact on hygrophila. In addition, a very damaging aecial rust fungus (Puccinia sp.) was collected. Although initial studies suggested that this rust could be the aecial stage of the heteroecious P. cacao, detailed cross-inoculation studies involving its gramineous primary host H. compressa are necessary to confirm its identity.

141

As part of the surveys to identify natural enemies, phytoparasitic nematodes present in the root zone of hygrophila also were extracted and identified. This study demonstrated for the first time the presence of phytoparasitic nematodes in the root zone of the invasive weed in its native and exotic ranges. Significantly higher densities of phytoparasitic nematodes were associated with hygrophila in India compared to

Florida. The high densities of phytoparasitic nematodes in the root zone of hygrophila suggest that this weed may act as an alternative host of some important crop pests.

Several studies have confirmed that climate matching between exotic and native ranges is crucial as climatic incompatibility can severely limit establishment and effectiveness of biological control agents. Furthermore, a paucity of specific records of hygrophila occurrences in its native range made it difficult to prioritize areas for future surveys. In this study, the Maximum Entropy Species Distribution Model (MaxEnt) was used to prioritize native areas in India and Bangladesh to conduct exploration for biological control agents. MaxEnt modeling is based on the assumption that climatic tolerance of a species primarily determines its distribution. A threshold dependent binomial test was used to verify the accuracy of the modeled distribution based on known occurrences of hygrophila. The results of these tests demonstrated that MaxEnt was very accurate in predicting the known distribution of hygrophila and included areas that are ecologically favorable for its growth. The modeled distribution predicted high suitability for the presence of hygrophila in the northeastern regions of India and in northern Bangladesh. Based on the results of this study, new areas were prioritized for conducting future surveys. These include areas in Meghalaya and Tripura along the

Bangladesh border, areas along the River Ganges in Northern Bihar and Utter Pradesh,

142

and places in Orissa across the River Mahanadi. Bangladesh was found to particularly suitable, as most regions of the country (particularly areas in the northern plains) were predicted to be suitable for growth of this weed. Because only a small area in

Mymensingh, Bangladesh was surveyed in this study, more extensive surveys are needed. Prioritized areas for future surveys include places in Dinajpur, Rangpur,

Jamalpur, Sylhet, as well as northern parts of Bogra and Rajsahi. This study confirmed that MaxEnt can be an effective tool in classical biological control for identifying climatically suitable native habits for foreign exploration studies. Overall, the results generated in this study confirmed that hygrophila is a suitable target for biological control.

143

APPENDIX A RECORDS OF HYGROPHILA POLYSPERMA COLLECTED FROM THE CENTRAL NATIONAL HERBARIUM, HOWRAH, INDIA

Sl. Latitude* Longitude* State Locality Information Date of collection Other information No.

Sandy bank of Sarbhanga 1 27.238433 94.109914 Assam 24 April,1957 Alt 4000ft river

Waterplant rooting in swampy mud 2. 25.925069 85.864617 Bihar Nipani village 19.12.1957 bank, flower bluish white

Himachal 3. 32.098561 76.266617 Kangra district June 1911 Alt 3000ft Pradesh

Madhya 144 4. 25.442581 78.566328 Bawar sapor near Jhanshi 09.12.1986 - Pradesh

Madhya Rooting in the sandy mud of the 5. 22.083969 82.155022 Bilaspur, Pasan Feb, 1972 Pradesh pond, flower bluish white

Madhya 6. 24.362489 79.209528 Sagor District, Hirapur tank 03.03.1960 Altitude 430m, Pradesh

Madhya 7. 21.757269 81.568011 Bastar district, near Kanker 06.02.1961 - Pradesh Annual herb with white roots growing Madhya 8. 22.744278 77.718789 Hosangabad, Bhains Dehi 22.07.1964 near water ditch, stem stripped little Pradesh hairy Madhya Under shrub, radial leaves, densely 9. - - Bilaspur, Korba 18.04.1965 Pradesh hairy & smaller, flower bluish white.

Madhya Small plants in moist places, flower 10. 24.529944 81.294814 Rewa district, Dabhoura (?) 27.02.1987 Pradesh purple

Madhya 11. 19.600431 81.670211 Bastar district, Kondagaon 19.11.1958 Altitude 767m., Semi aquatic plant Pradesh

Madhya Jubbalpore district, Road to Alt. 490m, white flowers with pink 12. 23.163686 79.9442 12.03.1962 Pradesh Goriah tinge

Flowers pale lilac, marsh plant, 13. 21.25 79.1 Maharashtra Nagpur, Koradi 05.12.1961 seeds many

14. - - Maharashtra - - -

15. 24.818786 93.883342 Manipur Lamphel 12.01.1954 -

Sambalpur district, 16. 21.464878 83.976861 Orissa 06-11-1986 Abundant Jailchhaka

145 Jagannath Sagar (on the 17. 20.241983 85.835 Orissa 14.12.1962 - bank)

18. 31.527981 75.91375 Punjab Hooshiarpur 1878 Collected by Dr. Aitchinson

Ascending herb, flowers whitish, and Bharatpur district, Ghana bird 19. 30.462667 76.376239 Rajasthan 27.03.1982 stamens 2. Soil moist clayey, not sanctuary common Annual herb in moist places up to 20. 23.832264 73.714742 Rajasthan Dungarpur district - 5cm flower white

21. 23.540544 74.44225 Rajasthan Uadipur district, Banswara 17.04.1963 Annual herb near pond

An erect, annual nearly 30cm high 22. 26.616717 73.906797 Rajasthan Samarania 18.05.1965 herb, fruiting common in stagnant water

Annual herb in moist places up to 23. Rajasthan Dungarpur, Marwa 13.04.1963 5cm flowers white Alt 500m, Prostrate spreading herb, rooting at the nodes, flowers violet or pale pink, petals ciliate at the 24. Rajasthan, Kolar tank side 19.04.1977 margins, capsule immature, green, mature brown, soil clayey, common in marshy places

25. 27.251656 88.458411 Sikkim Tista river bed 22.02.1910 Plains

26. 27.603406 88.464564 Sikkim Baqrakoti 22.02.1911 Plains

Uttar 27. 29.967108 77.544594 Saharanpur District - - Pradesh

146 Forming a mat on shallow stagnant Uttar 28. 27.573519 81.595067 Baharaich district Rupaidigha 11.03.1964 water on bank. Inflorescence axis

Pradesh leafy, flower pinkish Creeping rhizome on the surface of Uttar 29. 29.1484 82.563081 Mirjapur District, Kotwa 08.03.1970 wet alluvial soil, flower bluish white, Pradesh minute Spreading & rooting at nodes, Uttar alternate oblong ( 2- 25 cm) leaf, 30. 27.559039 78.651453 Allahabad: Dariaganj 19.11.1964 Pradesh axillary flower, growing in moist situation Uttar Annual herb, up to 9” inside mud, 31. 25.44325 81.828408 Sahore,Allahabad 09.03.1964 Pradesh flower minute white. Annual diffused herb on road side Uttar Baharaich, Rupaidigha, 32. 27.573519 81.595122 09.03.1964 ditch. Rooting at nodes, flower Pradesh Abdullaganj slightly violet Annual diffused herb in very moist Uttar Baharaich District, Katarian 33. - - 16.02.1965 soil, stem rooting at nodes, leaves Pradesh Ghat Mandi light green, flower light pink, minute.

Uttar 34. 26.754961 83.36245 Gorokhpur - Flora of upper Gangetic plains Pradesh

Uttar Erect to prostrate, growing near 35. 27.866731 81.495353 Baharaich district, Nanpara 13.12.1986 Pradesh fields.

Uttar 36. 27.127069 81.960839 Gonda District, Balpur 1.06.1986 Herbs, Flower bluish white Pradesh

Uttar Baharaich district, Annual up to 6m in sandy moist soil 37. 25.050389 83.213097 07.02.1965 Pradesh Rupaidigha, Chakia on the edge of forest. Fruits dried up. Diffused annual herb. Flowers bluish Uttar Baharaich district, Murtiha 38. - - 12.3.1964 white, hairy fruits dehiscent by 2 Pradesh gate lobes

39. 26.541156 88.672136 West Bengal Sukna, Terai Region 21.12.1914 -

147 Jalpaiguri District, NEC beat 40. 26.516153 88.468611 West Bengal of Jaldapara wildlife 05.12.1995 -

sanctuary, BD - 1 Jalpaiguri district, Terrestrial or aquatic herb leaves 41. 26.522019 88.717414 West Bengal Chapramoni Wildlife - opposite, terminal spike, corolla sanctuary bilobed Jaldapara Wildlife sanctuary, 42. - - West Bengal 19.12.1985 BD – 2, NEC beat

Hooghly District, 43. 22.904194 88.374711 West Bengal 03.12.1967 Common herb, white flower Champadanga (riverside) A small procumbent weed with small 44. 23.632736 87.529886 West Bengal Birbhum district, Ilambazar - flowers by the roadside of an dried jheel, very common Jalpaiguri district, Buxa, 45. 26.678708 89.741519 West Bengal 22.11.1975 - Nimati range

Mecheda, Purba Medinipur 46. 26.349492 87.896731 West Bengal - - district Small much branched prostrate herb Cooch Behar District, Torsa 47. 26.320814 89.463289 West Bengal 03.03.1984 with small leaves modes swollen, river bed in sand stem pinkish, fruit dehiscent

48. 22.904125 88.374711 West Bengal Hooghly district, Nalikul 27.01.1963 -

49. 23.232431 87.863633 West Bengal Bardhaman district, Lakhodih 04.05.1960 -

50. 22.869831 87.953772 West Bengal Rasulpur village - -

51. 22.420878 87.326089 West Bengal Midnapur district, Chilkighar - -

148 52. - - West Bengal Medinipur district - Common

Bandel, Polba village, West 53. 22.918975 88.402189 West Bengal - - Bengal

Hooghly district 54. 22.834742 87.979028 West Bengal 0.12.1967 - Champadanga (River side)

Gopalpur, Bishnupur, 55. 23.2347 87.072381 West Bengal 14.01.1965 - Bankura district

56. 26.526808 88.738117 West Bengal Bhutan ghat, Jalpaiguri 02-12-1975 Alt 600m

Stamens 2.4mm, corolla 8.5mm, 57. - - West Bengal Moshat 21.04.1968 calyx, 5mm, fruit 6.5mm, gynoecium 3.3mm

58. West Bengal Jalpaiguri district, Balapara 06.12.1975 -

24 Parganas (s), Jagaddal 59. 22.45495 88.418456 West Bengal 30.04.1963 - village

Bankura District, Abantika 60. 23.079767 87.328536 West Bengal 14.01.1965 - Bishnupur

61. 22.650753 88.265833 West Bengal Howrah district, Dumjoor - -

62. 22.580036 88.255667 West Bengal Mourigram, Howrah district, - -

63. West Bengal Burdwan district, Lakhodih 04.05.1960 -

149 64. 23.338275 86.359461 West Bengal Purulia district, Saheb Bandh 10.03.1964 -

*Note that latitudes/longitudes were generated using head-up digitizing technique

APPENDIX B ROYAL BOTANIC GARDENS, KEW, UNITED KINGDOM HERBARIUM RECORDS FOR HYGROPHYLIA POLYSPERMA IN THE OLD WORLD

Site details Evidence of natural Habitat and plant details Miscellaneous (Date of collection) enemy damage Two plants, one with small Punjab, tropical Consistent leaf leaves (2 cm long), Flora of British India 1000feet spot, pathogen and compact plant. One with 1 v4 page 406 (24 May1848) insect holes with larger leaves (4cm long)

rim Seed evident curved (shattered) 2 Bengal Herbarium Kewense 8 Feb 1854 3 Seed Collected by Jr

Drummond 2 specimen sheets, one Herbarium Kewense India, Punjab 4 with large and one with Collected by Jr (1886) small leaves Drummond Herbarium Kewense India, Punjab Collected by Jr 5 Salt range Drummond (1887) Punjab province of Pakistan Collected by A.N. Nurpur, Kangra Wet ground, flowers and Parker; Internet 6 district, Punjab; 600m shattered seed pods search: Himachal (17 Apr 1938) Pradesh, India Flowers pale-lilac, Internet search: Flora of Nagpur, 7 submerged leaves filmy, Maharashtra state, Mansar seeds 20 or more central/west India In flower. Small decumbent Kemetipanga, Leaf minor and herb with simple opposite taken for study daspalla state stem minor exstipulate. Entire leaves at Kew botanical 8 700 feet (outside) significant and blue flowers in terminal gardens; Internet (17 Dec 1937) damage bracteates spikes. In paddy search: Orissa

field. Saharpulieir 9 Western Himalaya, (19 Mar 1897) Sikkim. 10 Tropical (Pre 1867) In bottom of dried-up tank. Nr. Mochibahal on Leaves look more main – elongated than other Herb. Mooney 11 Angul road, specimens (could be Called H. serphyllum Redhakhol, Orissa. alligator weed put in wrong (14/19 May 1949) place in herb?) In paddy field growing with Nutar, crop of Mung. Prostrate or state. procumbent herb with many 12 Called H. serphyllum 300 feet stems from stout rootstock. (15 Mar 1939) Flowers: purple, in auxiliary whorls calyx 5 lobed

150

Junagarh, Sambalpur, Leaf minor, Gregarious, in damp 13 Herb. Mooney Orissa; 700 feet Mildew? ground 14 Dec 1948 Western Himalaya. 14 Duthie (April 1898) There is a Moradabad Moradabad in Uttar 15 Malacca Pradesh; Malacca is (March 1844) a state in Malaysia! Lunbhet, Bamra state, 16 Sambalpur, Orissa. Moist depression in pasture Herb. Mooney (13 Nov 1940) Moradabad Internet search: Uttar (upper Ganges plain) 17 Tropical Pradesh, India 1000 feet

(March 1844) Sambalpur, Narsinghpur state, Leaf spot, insect feeding on 18 Herb. Mooney Orissa; 200 feet leaves (frass) (23 Feb 1941) Bhira range, Western Internet search: India Himalaya, upper - Nepal border, Uttar 19 Gangetic plain, Kheri Tiny stunted plants Pradesh, India (Oudh), Dudhwa, (Dudhwa national (8 Apr 1898) park) Gorakhpur, Western Internet search: utter Himalaya, upper 20 Elongated plants Pradesh, north east Gangetic plain India (27 Apr 1898) Insect leaf damage Type ex Malacca Could be in India, / lacking leaves 21 (Malaysia?) Malacca could be a shot holes (1845) reference to person? (chysomelid) Fairly common herb in Lower Burma, Pegu. 22 damp places. Found near (Jan 1948) edge of water Assam, Lakhimpur, Straggling herb 23 Badati. Flowers bluish-white (Dec 1946) Rangoon, Concan 24 (1867) Chittagong, Internet search: 25 Tropical habitat 100 ft (1850) Bangladesh Himalayan herbarium Internet search: Insect holes and 26 Kosilla; 1500 feet Kosilla possibly river leaf spot (1850) in Bangladesh Dharwar Internet search: 27 Insect holes (Nov 1851) Karnataka India Talod, Ahmadabad Internet search: India 28 Marshy ground (Gujarat) (Nov. 1914) near to Pakistan Pattarium, ?Purwean No internet search 29 district; (Dec 1918) match Gundaepur, Punjab, Mildew Wet place (Feb 1917)

151

Naga hills, Manipur 30 (Dec 1907) Rajpur, Bengal 31 (March 1904) Upper Gangetic plain, 32 Gonda district. Uttar Pradesh (Apr 1898) Argadata No internet search 33 (Feb 1884) match Leaf minor. Khandala, Western Rust pustules 34 Ghats, (upper leaf surface) (Mar 1937) pale tan/brown Type: two collections? 35 1830, East India 1841, upper Assam Sri Lanka: Habarana, Under shade along bank of 36 Polonnaruwa tank; common (Apr 1993) Sri Lanka, Leaf minor. 37 Trincomalee district, Possible rust on Uppuveli (Jul 1993) lower leaf. Sri Lanka, Inamalana 38 New to Ceylon (Jun 1932) Could have been Sri Lanka, introduced. Locally Mannampitiya, Annual, white common, in soggy clayey 39 Gengadu Villu. Collected by Cramer flowers soil among short grass Pononnaruwa. 60m along marshy borders of (11 Apr 1987) villu. Taiwan, Tckow plain, Insect holes, leaf 40 Blue flower, in ditch (Apr 1895) spot Vietnam (Tonkin) 41 (Nov 1887)

Collected by: Carol Ellison and Harry Evans, CABI-Europe, UK; November 2007

152

APPENDIX C DESCRIPTIONS OF FIELD COLLECTION SITES IN INDIA

EC DO Date Location Geoposition Elevation (ft) Ph Habitat Description (mS/cm) (mg/L) Majh Baghmaro, 26.735272, Road side small pool. Huge growth of 16 Sep 08 Biswanath Chariali, 93.169239 hygrophila. Leaf damage observed. Tezpur Assam Gingia Lalpani, Tezpur 26.818714, Marshy land with stagnant rain water. 16 Sep 08 Assam 93.276361 Population level low, new growth Majh Baghmaro, 26.735272, Same spot visited earlier. Diseased 17 Sep 08 Biswanath Chariali, 93.169239 plant parts collected Tezpur Assam Genetic analysis samples collected. Hawajan, Lakhimpur, 26.894833, 17 Sep 08 198.8 Newly emerged population. Not much Assam 93.782083 leaf damage observed Betbari, Sibsagar 27.062056, Road side population. Frequented by

153 18 Sep 08 342.5 District, Assam 94.668956 cattle, quite disturbed

Malowakhat, Jorhat, Plants found growing in irrigation 19 Sep 08 - - Assam channel. Not much leaf damage Drainage canal with flowing water surrounded by paddy fields. Botalikhusa, Jorhat, 26.768972, 19 Sep 08 213.5 Submerged population of hygrophila. Assam 94.118417 Leaf cutter damage observed and collected. At the bank of river Brahmaputra, 26.862667. Submerged population. Synclita like 19 Sep 08 Nimati, Jorhat, Assam 221.9 94.235997 underwater leaf cutter observed. Later identified as Parapoynx bilinealis Betbari, Sibsagar Road side population. Rust infestation 20 Sep 08 District, Assam found for the first time on hygrophila. Very dry drainage canal with Rani Gate, beside VIP 26.085692, 14 Oct 08 120.4 cemented base. Mat like growth of road, Guwahati, Assam 91.616222 hygrophila.

National Highway 37 on the way from Dharamtul ( NH 37), 26.164917, 14 Oct 08 172.3 6.3 019 7.5 Guwahati to Tezpur. Completely Assam 92.356111 submerged population. Majhbargaon, Tezpur Same road side population visited 26.735256, 15 Oct 08 Biswanath Chariali, 228.3 6.3 089 7.5 earlier. More green larvae collected. 93.169236 Assam Soil sample collected Kadamani Bill, 26.71725, Big water body with huge terrestrial 15 Oct 08 Biswanath Chariali, 206.5 6.1 093 6.6 93.129056 population. Sweep sampling done Assam Lake Kumulia, Close to the earlier water body, 26.685883, 15 Oct 08 Biswanath Chariali, 185.1 6.1 055 7.9 Monoculture of hygrophila. Leaf 93.134667 Assam damage observed. 26.862667, Submerged population, many 16 Oct 08 Nimati, Jorhat, Assam 251 6.3 155 7.9 94.236006 underwater leaf cutter larvae collected Batali Khusa, Jorhat, 26.768972, 16 Oct 08 213.5 6.0 094 8.3 Black larva damage was observed. Assam 94.118417 Small rivulet very near to tea garden. Outside Gibbon 26.669114, Underwater population of hygrophila. 16 Oct 08 National Park, Jorhat, 327.4 5.8 069 6.5 154 94.351714 Leaf cutter damage observed and Assam collected. Small canal with very murky water. Rajmai Near Dimu, 27.100639, 17 Oct 08 171.3 5.8 023 7.8 Hygrophila plants growing on the Assam 94.702353 bank. Betbari, Sibsagar 27.062056, The rust site visited earlier. Collected 18 Oct 08 342.5 5.7 029 7.8 District, Assam 94.668956 many rusted plants for inoculation Road side submerged paddy field. Near Agunkhaki 22.177861, 6 Nov 08 14.4 6.6 323 7.9 Hygrophila is growing with other Samsan, Nimpith 88.443833 aquatic plants Narendrapur,24 22.419406, Big lake, plants are growing with 8 Nov 08 Parganas (S), West 7.6 806 9.5 88.393556 water hyacinth. Bengal Narendrapur, 24 22.419175, Abandoned Brick field. Very disturbed 8 Nov 08 Parganas (s), West 88.36915 habitat. Only soil samples collected Bengal Nandapara on Majdeo Road side population. Big water body, Road near Bhimpur, 23.42775, 12 Nov 08 - 4.0 7.5 105 8.3 plants growing mixed with other Krishna Nagar, West 88.550167 aquatic plants Bengal

Nandapara on Majdeo 23.419306, Road side population, population level 12 Nov 08 Road, Nearby place, 8.7 88.568986 low West Bengal Bhimpur, Krisnanagar, 23.432972, Big ponds, submerged population. 12 Nov 08 28.1 6.9 321 8.3 West Bengal 88.653833 Collected for Berlese funnel extraction Bahadurpur, Near Bank 23.4235, River side population, leaf damage 12 Nov 08 of River Jalangi, West 106.8 6.6 316 8.5 88.470722 observed Bengal Hygrophila growing along the bank of Muratpur, Kalyani, 13 Nov 08 the river. Leaf miner damage Nadia, West Bengal observed. Green larvae also present Balapara Bandh, Near Very disturbed habitat. Lots of dirt on 26.551678, 19 Nov 08 Teesta Bridge, 239.2 7.2 178 8.4 leaf. Plants collected for Berlese 88.734611 Jalpaiguri, West Bengal extraction Between Damani More Plants growing along the paddy field. 26.561694, 19 Nov 08 and Mynaguri Road, 231.2 6.7 105 8.4 Rust infestation observed. Green Leaf 88.768997 Jalpaiguri, West Bengal cutting semilooper also observed 26.552017, Very near to Teesta river, samples for 19 Nov 08 Side of Teesta Bridge 256.5 - - - 88.736092 genetic analysis collected 155 Dangapara Engineering 26.549583, Green larvae semilooper observed 19 Nov 08 230.9 6.4 87 9.4

College More 88.702722 and collected Naya Nadi Bridge on Huge population of hygrophila Falakata to Pundibari 26.401997, growing under a bridge. Flowing 20 Nov 08 Road, Near Pundibari, 125.4 6.8 510 9.4 89.351919 water also present. Sweep sampling Cooch Behar, West done. Rust infection also observed Bengal Near to Malda railway station. Ghorapir at Kazigram, 25.003556, 22 Nov 08 22.4 6.9 413 9.0 Population level low. Samples for Malda, West Bengal 88.102767 nematode and soil collected Huge population of hygrophila growing beside a canal with flowing Suzapur, Kaliachak – 1, 24.922083, 22 Nov 08 21.6 6.9 514 10.3 water. Rusted plants were collected. Malda, West Bengal 88.085014 Soil samples collected for nematode extraction

APPENDIX D DESCRIPTIONS OF FIELD COLLECTION SITES IN BANGLADESH

Elevation EC DO Habitat Description Date Sites Latitude Longitude pH (ft) (mS/cm) (mg/L) Roadside fallow land. Rust Sikarikanda, infection, Aphid with virus 11 Oct 09 24.710194 90.410944 118.8 8.5 300 12.5 Mymansingh infection observed. Soil collected for nematode extraction Road side shallow area. Leaf Churkhai, 11 Oct 09 24.666556 90.402556 37.1 6.8 100 7.2 damage not observed. Aphid Mymansingh population with virus infection Nematode sample and genetic Trishal, analysis samples collected. 11 Oct 09 24.544611 90.379833 3.18 6.5 100 8.7 Mymansingh Aphid infestation with virus infection observed. 156 Shallow lake and marshy area 12 Oct 09 Galakanda Union, with submerged and emerged 24.848444 90.597667 23.62 6.5 100 10.1 Netrokona hygrophila. Leaf cutting caterpillar observed. Road side marsh. Large area Krishnakali, 12 Oct 09 24.899806 90.751361 - 4.2 6.7 100 8.6 with huge hygrophila population. Netrokona Severely infected with rust. Dawapara, Small area with thin population. 12 Oct 09 24.767722 90.752306 - 12.8 7.9 100 8.9 Netrokona Rust infection observed. Rajpur, Kendua, Very low land with stagnant 12 Oct 09 24.653306 90.820111 6.5 6.8 100 9.8 Netrokona water. Population level low. Precis almana larva, Trachys 12 Oct 09 Baruigram, 24.573222 90.730528 9.2 6.9 100 10.8 spp. adult collected. Sample for Mymansingh genetic analysis collected BAU, In front of Marshy area. Clumps of 13 Oct 09 BINA Guest 24.725306 90.432889 29.9 6.4 300 10.2 hygrophila. Sample for genetic House analysis were collected. Hygrophila population with rust Sutiakhlai, 13 Oct 09 24.697556 90.455667 2.6 7.7 900 8.5 infection observed. Marsh, with Mymansingh stagnant water.

Big population of hygrophila as Sutiakhlai, Bank found at the bank of the river 13 Oct 09 of Brahmaputra, 24.684139 90.458694 14.4 6.6 200 9.4 Brahmaputra. The green Noctuid Mymansingh larva collected. Rust infection also observed. Road side marshy field. 13 Oct 09 Kalibazar, Trishal 24.628556 90.466389 10.82 6.7 100 8.7 Population level low. Not much leaf damage observed. Muddy drainage basin, with clumps of hygrophila. Leaf Biarta, Court para, 13 Oct 09 24.615333 90.49225 12.1 6.7 100 8.5 damage observed. Samples Mymansingh collected for Further observations

157

LIST OF REFERENCES

Ahmad, K.J., 1976. Epidermal Studies in some Species of Hygrophila and Dyschoriste Acanthaceae. Journal of the Indian Botanical Society 55, 41-52.

Alagarswamy, G., Boote, K., Allen Jr, L., Jones, J., 2006. Evaluating the CROPGRO- soybean model ability to simulate photosynthesis response to carbon dioxide levels. Agronomy Journal 98, 34 - 42.

Alexandris, S., Stricevic, R., Petkovic, S., 2008. Comparative analysis of reference evapotranspiration from the surface of rainfed grass in central Serbia, calculated by six empirical methods against the Penman-Monteith formula. European Water 21/22, 17-28.

Analla, M., 1998. Model validation through linear regression fit to actual versus predicted values. Agricultural Systematics 57, 115 - 119.

Anderson, R.P., Gomez-Laverde, M., Peterson, A.T., 2002. Geographical distributions of spiny pocket mice in South America: insights from predictive models. Global Ecology and Biogeography 11, 131-141.

Anderson, R.P., Lew, D., Peterson, A.T., 2003. Evaluating predictive models of species' distributions: criteria for selecting optimal models. Ecological Modelling 162, 211- 232.

Anderson, R.P., Martinez-Meyer, E., 2004. Modeling species' geographic distributions for preliminary conservation assessments: an implementation with the spiny pocket mice (Heteromys) of Ecuador. Biological Conservation 116, 167-179.

Anderson, R.V., Potter, J.W., Nickle, W.R, 1991. Stunt nematodes: Tylenchorhynchus, Merlinius, and related genera, in: W. R. Nickle (Ed.), Manual of Agricultural Nematology., MarCel Dekker Inc., New York. pp. 529-586.

Angerstein, M.B., Lemke, D.E., 1994. First records of the aquatic weed Hygrophila polysperma (Acanthaceae) from Texas. Sida Contributions to Botany 16, 365-371.

Ascunce, M.S., Yang, C.C., Oakey, J., Calcaterra, L., Wu, W.J., Shih, C.J., Goudet, J., Ross, K.G., Shoemaker, D.W., 2011. Global Invasion History of the Fire Ant Solenopsis invicta. Science 331, 1066-1068.

Australian Weeds Committee, 2010. Noxious weed list for Australian states and territories. Australian Weeds Committee, pp. 1 - 25.

Bailey, P., 2007. Pests of Field Crops and Pastures, Identification and Control. CSIRO, Collingwood, Victoria, Australia.

158

Balciunas, J. K., Burrows, D. W., 1996. Distribution, abundance and field host-range of Hydrellia balciunasi Bock (Diptera: Ephydridae) a biological control agent for the aquatic weed Hydrilla verticillata (Lf) Royle. Australian Journal of Entomology. 35, 125-130.

Baldwin, I.T., 1990. Herbivory simulations in ecological research. Trends in Ecology & Evolution 5, 91-93.

Baldwin, R.A., 2009. Use of Maximum Entropy Modeling in Wildlife Research. Entropy 11, 854-866.

Balsbaugh, E.U., 1970. Review of the Genus Paria (Coleoptera: Chrysomelidae) of North America. Annals of the Entomological Society of America 63, 453-460.

Bartel, R.A., Sexton, J.O., 2009. Monitoring habitat dynamics for rare and endangered species using satellite images and niche-based models. Ecography 32, 888-896.

Baudoin, A., Abad, R., Kok, L., Bruckart, W., 1993. Field evaluation of Puccinia carduorum for biological control of musk thistle. Biological Control 3, 53-60.

Beaumont, L.J., Gallagher, R.V., Thuiller, W., Downey, P.O., Leishman, M.R., Hughes, L., 2009. Different climatic envelopes among invasive populations may lead to underestimations of current and future biological invasions. Diversity and Distributions 15, 409 - 420.

Bellamy, C.L., 2008. A World Catalogue and Bibliography of the Jewel Beetles (Coleoptera: ). Pensoft.

Belsky, A.J., 1987. The Effects of Grazing: Confounding of Ecosystem, Community, and Organism Scales. The American Naturalist 129, 777 - 783.

Bernard, E.C., D.P. Schmitt. 2005. Nematode assemblages in native plant communities of Molokai, Hawaii. Journal of Nematology 37:242-248.

Boote, K., Jones, J., Mishoe, J., Berger, R., 1983. Coupling pests to crop growth simulators to predict yield reductions. Phytopathology 73, 1581-1587.

Botts, P.S., Lawrence, J.M., Witz, B.W., Kovach, C.W., 1990. Plasticity in Morphology, Proximate Composition, and Energy Content of Hygrophila polysperma (Roxb) Anders. Aquatic Botany 36, 207-214.

Boubli, J. P., Lima, M. G. D., 2009. Modeling the geographical distribution and fundamental niches of Cacajao spp. and Chiropotes israelita in northwestern Amazonia via a maximum entropy algorithm. International Journal of Primatology. 30, 217-228.

Brunel, S., 2009. Pathway analysis: aquatic plants imported in 10 EPPO countries. EPPO Bulletin 39, 201-213.

159

Buckingham, G. R., Bennett, C. A., 1996. Laboratory biology of an immigrant Asian moth, Parapoynx diminutalis (Lepidoptera: Pyralidae), on Hydrilla verticillata (Hydrocharitacea). Florida Entomologist. 79, 353-363.

Buckingham, G.R., 1994. Biological control of aquatic weeds. Pest management in the subtropics: Biological control: A Florida perspective, 413-480.

Buckingham, G.R., Bennett, C.A., 1989. Laboratory host range of Parapoynx diminutalis (Lepidoptera: Pyralidae), an Asian aquatic moth adventive in Florida and Panama on Hydrilla verticillata (Hydrocharitaceae). Environmental Entomology 18, 526- 530.

Buermann, W., Saatchi, S., Smith, T.B., Zutta, B.R., Chaves, J.A., Mila, B., Graham, C.H., 2008. Predicting species distributions across the Amazonian and Andean regions using remote sensing data. Journal of Biogeography 35, 1160-1176.

Byrne, M.J., Currin, S., Hill, M.P., 2002. The influence of climate on the establishment and success of the biocontrol agent Gratiana spadicea, released on Solanum sisymbriifolium in . Biological Control 24, 128-134.

CABI. 2005 Crop Protection Compendium, 2005 Edition., CAB International, Wallingford, UK.

Campbell, C., McCaffrey, J., 1991. Population trends, seasonal phenology, and impact of Chrysolina quadrigemina, C. hyperici (Coleoptera: Chrysomelidae), and Agrilus hyperici (Coleoptera: Buprestidae) associated with in northern Idaho. Environmental Entomology 20, 303-315.

Capinera, J.L., 1999. Fall armyworm, Spodoptera frugiperda (J.E. Smith) (Insecta: Lepidoptera: Noctuidae). Featured Creatures from the Entomology and Nematology Department, Florida Cooperative Extension, Gainesville, Florida. http://entnemdept.ufl.edu/creatures/field/fall_armyworm.htm

Capinera, J.L., Roltsch, W.J., 1980. Response of wheat seedlings to actual and simulated migratory grasshopper defoliation. Journal of Economic Entomology 73, 258 - 261.

Carr, G.M., Duthie, H.C., Taylor, W.D., 1997. Models of aquatic plant productivity: a review of the factors that influence growth. Aquatic Botany 59, 195-215.

Cassani, J.R., 1996. Managing aquatic vegetation with grass carp. A guide for water resource manager. American Fisheries Society, Introduced Fish Section (Bethesda, Md), Bethesda, Md.

Castoe, T.A., Poole, A.W., Gu, W., Jason de Koning, A., Daza, J.M., Smith, E.N., Pollock, D.D., 2010. Rapid identification of thousands of copperhead snake (Agkistrodon contortrix) microsatellite loci from modest amounts of 454 shotgun genome sequence. Molecular Ecology Resources 10, 341-347.

160

Center, T.D., Dray, F.A.J., Jubinsky, G.P., Grodowitz, M.J., 2002. Insects and other arthropods that feed on aquatic and wetland plants. United States Department of Agriculture, Agricultural Research Service, Technical Bulletin 1870.

Charudattan, R., 2001. Biological control of weeds by means of plant pathogens: significance for integrated weed management in modern agro-ecology. Biocontrol 46, 229-260.

Chen, F., Song, S., Wu, C., 2006. A review of the genus Parapoynx Hübner in China (Lepidoptera: Pyralidae: Acentropinae). Aquatic Insects 28, 291-303.

Choong, M., 1996. What makes a leaf tough and how this affects the pattern of Castanopsis fissa leaf consumption by caterpillars. Functional Ecology 10, 668- 674.

Coile, N.C., 1995. Common plants of Florida‟s aquatic plant industry. Florida Department of Agriculture and Consumer Services, Division of Plant Industry.

Colautti, R.I., A. Ricciardi, I.A. Grigorovich, H.J. MacIsaac. 2004. Is invasion success explained by the enemy release hypothesis? Ecology Letters 7:721-733.

Cook, C., Cook, C.D.K., 1996. Aquatic and wetland plants of India. Oxford University Press Inc, New York.

Corder, G.W., D.I. Foreman. 2009. Nonparametric statistics for non-statisticians: a step- by-step approach John Wiley & Sons Inc. Hoboken, New Jersey: ix+245p.

Cox, G.W., 1999. Alien species in North America and Hawaii: impacts on natural ecosystems. Island Pr.

Creed, R. P., Sheldon, S. P., 1993. The effect of feeding by a North American weevil, Euhrychiopsis lecontei, on Eurasian watermilfoil (Myriophyllum spicatum). Aquatic Botany. 45, 245-256.

Croxton, M.D., Andreu, M.A., Williams, D.A., Overholt, W.A., Smith, J.A., 2011. Geographic Origins and Genetic Diversity of Air-Potato (Dioscorea bulbifera) in Florida. Invasive Plant Science and Management 4, 22-30.

Cuda, J. P., Coon, B.R., Dao, Y.M., Center, T.D., 2002. Biology and laboratory rearing of Cricotopus lebetis (Diptera: Chironomidae), a natural enemy of the aquatic weed Hydrilla (Hydrocharitaceae). Annals of the Entomological Society of America. 95, 587-596.

Cuda, J.P., Charudattan, R., Grodowitz, M.J., Newman, R.M., Shearer, J.F., Tamayo, M.L., Villegas, B., 2008. Recent advances in biological control of submersed aquatic weeds. Journal of Aquatic Plant Management 46, 15-32.

161

Cuda, J.P., Sutton, D.L., 2000. Is the aquatic weed hygrophila, Hygrophila polysperma (Polemoniales: Acanthaceae), a suitable target for classical biological control? Proceedings of the X International Symposium on Biological Control of Weeds, Bozeman, Montana, USA, 4-14 July, 1999, 337-348.

Cullen, J., Kable, P., Catt, M., 1973. Epidemic spread of a rust imported for biological control. Nature 244, 462-464.

Cullings, K.W., 1992. Design and testing of a plant-specific PCR primer for ecological and evolutionary studies. Molecular ecology 1, 233-240.

Dai, W., Viraktamath, C., Zhang, Y., Webb, M., 2009. A review of the genus Scaphotettix Matsumura (Hemiptera: Cicadellidae: Deltocephalinae), with description of a new genus. Zoological science 26, 656-663.

Dash, P.C., Viraktamath, C., 1997. New species of grass feeding hecaline leafhopper genera Glossocratus and Hecalus (Hemiptera: Cicadellidae) from India.

Davis, A.S., Landis, D.A., Nuzzo, V., Blossey, B., Gerber, E., Hinz, H.L., 2006. Demographic models inform selection of biocontrol agents for garlic mustard (Alliaria petiolata). Ecological Applications 16, 2399-2410.

Davis, R., Webster, T., Brenneman, T., 2006. Host status of tropical spiderwort (Commelina benghalensis) for nematodes. Weed Science 54:1137-1141.

DeLoach, C.J., Cordo, H.A., 1976. Life Cycle and Biology of Neochetina bruchi, a weevil attacking water hyacinth in Argentina, with notes on N. eichhorniae. Annals of the Entomological Society of America 69, 643-652.

Dennill, G.B., Gordon, A.J., 1990. Climate-related differences in the efficacy of the Australian gall wasp (Hymenoptera, Pteromalidae) released for the control of Acacia longifolia in South Africa. Environmental Entomology 19, 130-136.

DeVaney, S.C., McNyset, K.M., Williams, J.B., Peterson, A.T., Wiley, E.O., 2009. A Tale of Four "Carp": Invasion Potential and Ecological Niche Modeling. PLoS ONE 4, e5451.

Dhileepan, K., Lockett, C., Robinson, M., Pukallus, K., 2009. Prioritising potential guilds of specialist herbivores as biological control agents for prickly acacia through simulated herbivory. Annals of Applied Biology 154, 97-105.

Dhileepan, K., Senaratne, K., Raghu, S., 2006. A systematic approach to biological control agent exploration and prioritisation for prickly acacia (Acacia nilotica ssp indica). Australian Journal of Entomology 45, 303-307.

Distant, W., 1902. Rhynchotal notes. XIII. Heteroptera: Families Tingididae, Phymatidae and Aradidae. Annals And Magazine of Natural History 9, 353-362.

162

Distant, W., 1909. New Oriental Tingididae. Annales de la Socié té Entomologique de Belgique 53, 113-123.

Doyle, J.J., Doyle, J.L., 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochemistry Bulletin 19, 11-15.

Doyle, R.D., Francis, M.D., Smart, R.M., 2003. Interference competition between Ludwigia repens and Hygrophila polysperma: two morphologically similar aquatic plant species. Aquatic Botany 77, 223-234.

Eckert, C.G., Lui, K., Bronson, K., Corradini, P., Bruneau, A., 2003. Population genetic consequences of extreme variation in sexual and clonal reproduction in an aquatic plant. Molecular ecology 12, 331-344.

EDDMaps, 2010. Early Detection and Distribution Mapping System - Distribution of Hygrophila polysperma in the United States. http://www.eddmaps.org/.

Elbir, T., 2003. Comparison of model predictions with the data of an urban air quality monitoring network in Izmir, Turkey. Atmospheric Environment 37, 2149 – 2157.

Elith, J., Graham, C.H., Anderson, R.P., Dudik, M., Ferrier, S., Guisan, A., Hijmans, R.J., Huettmann, F., Leathwick, J.R., Lehmann, A., Li, J., Lohmann, L.G., Loiselle, B.A., Manion, G., Moritz, C., Nakamura, M., Nakazawa, Y., Overton, J.M., Peterson, A.T., Phillips, S.J., Richardson, K., Scachetti-Pereira, R., Schapire, R.E., Soberon, J., Williams, S., Wisz, M.S., Zimmermann, N.E., 2006. Novel methods improve prediction of species' distributions from occurrence data. Ecography 29, 129-151.

Elith, J., Leathwick, J. R., 2009. Species distribution models: Ecological explanation and prediction across space and time. Annual Review of Ecology, Evolution, and Systematics. 40, 677 - 697.

Ellison, C.A., Evans, H.C., Djeddour, D.H., Thomas, S.E., 2008. Biology and host range of the rust fungus Puccinia spegazzinii: A new classical biological control agent for the invasive, alien weed Mikania micrantha in Asia. Biological Control 45, 133-145.

Ellstrand, N.C., Roose, M.L., 1987. Patterns of genotypic diversity in clonal plant species. American Journal of Botany 74, 123-131.

Evanno, G., Regnaut, S., Goudet, J., 2005. Detecting the number of clusters of individuals using the software STRUCTURE: a simulation study. Molecular ecology 14, 2611-2620.

Evans, H.C., Ellison, C.A., 2005. The biology and taxonomy of rust fungi associated with the neotropical vine Mikania micrantha, a major invasive weed in Asia. Mycologia 97, 935.

163

Ferriter, A., Doren, B., Goodyear, C., Thayer, D., Burch, J., Toth, L., Bodle, M., Lane, J., Schmitz, D., Snow, S., Langeland, K., 2006. Chapter 9: The Status of Nonindigenous Species in the South Florida Environment. 2006 South Florida Environmental Report, Executive Summary. South Florida Water management District, Florida Department of Environmental protection, pp. 1-52.

Fiaboe, K. K. M., Fonseca, R.L., de Moraes, G.J., Gol, C., Knapp, M., 2006. Identification of priority areas in South America for exploration of natural enemies for classical biological control of Tetranychus evansi (Acari: Tetranychidae) in Africa. Biological Control. 38, 373-379.

FLDEP, 2007. Status of the Aquatic Plant Maintenance Program in Florida Public Waters; Annual Report for Fiscal Year 2006 - 2007. Florida Department of Environmental Protection, Tallahassee, Florida.

FLEPPC. 2009 List of Invasive Plant Species, Florida Exotic Pest Plant Council. Available at http://www.fleppc.org/list/09list.htm

Forno, I.W., Julien, M.H., 2000. Success in biological control of aquatic weeds by arthropods. In: Gurr, G., Wratten, S., (Eds.), Biological Control: Measures of Success. Kluwer Academic Publishers, Netherlands, pp. 159-187.

Fortuner, R., W.R. Nickle. 1991. The Hoplolaiminae. In: W. R. Nickle (Ed.), Manual of Agricultural Nematology, MarCel Dekker Inc., New York. pp. 669-719.

Freedman, J.E., Grodowitz, M.J., Swindle, R., Nachtrieb, J.G., 2007. Potential Use of Native and Naturalized Insect Herbivores and Fungal Pathogens of Aquatic and Wetland Plants (ERDC/EL TR-07-11 ). U.S. Army Corps of Engineers, Washington, DC.

Friedli, J., Bacher, S., 2001. Direct and indirect effects of a shoot-base boring weevil and plant competition on the performance of creeping thistle, Cirsium arvense. Biological Control 22, 219-226.

Gadd, M.E., Young, T.P., Palmer, T.M., 2001. Effects of simulated shoot and leaf herbivory on vegetative growth and plant defense in Acacia drepanolobium. Oikos 92, 515 - 521.

GBIF, 2011. Global Information Facility, http://data.gbif.org/species/13192166/, Accessed 06/April/2011

Geiger, J.H., Pratt, P.D., Koptur, S., 2010. Breeding System of Ruellia succulenta Small (Acanthaceae). Castanea 75, 454-460.

Genung, W.G., Mead, F.W., 1969. Leafhopper populations (Homoptera: Cicadellidae) on five pasture grasses in the Florida Everglades. The Florida Entomologist 52, 165-170.

164

Ghate, H., Borowiec, L., Rane, N., Ranade, S., Pandit, S., 2003. Tortoise beetles and their host plants from Pune (Maharashtra State, India) and nearby places (Coleoptera: Chrysomelidae: Cassidinae). Genus 14, 519-539.

Godfray, H.C.J., Waage, J.K., 1991. Predictive Modeling in Biological Control - The Mango Mealy Bug (Rastrococcus invadens) and its parasitoids. Journal of Applied Ecology 28, 434-453.

Goeden, R.D., 1983. Critique and revision of Harris' scoring system for the selection of insect agents in the biological control of weeds. Protection Ecology 5, 287-301.

Goolsby, J.A., De Barro, P.J., Makinson, J.R., Pemberton, R.W., Martley, D.M., Frohlich, D.R., 2006. Matching the origin of an invasive weed for selection of a herbivore haplotype for a biological control programme. Molecular ecology 15, 287-297.

Goolsby, J.A., Wright, A.D., Pemberton, R.W., 2003. Exploratory surveys in Australia and Asia for natural enemies of Old World climbing fern, Lygodium microphyllum: Lygodiaceae. Biological Control 28, 33-46.

Grime, J.P. 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. The American Naturalist 111:1169-1194.

Grinnell, J., 1917. The niche-relationships of the California Thrasher. The Auk 34, 427 - 433.

Grodowitz, M.J., Smart, M., Doyle, R.D., Owens, C.S., Bare, R., Snell, C., Freedman, J., Jones, H., 2004. Hydrellia pakistanae and H. balciunasi, insect biological control agents of hydrilla: boon or bust? Proceedings of the XI International Symposium on Biological Control of Weeds, Canberra, Australia, 27 April - 2 May, 2003, 529- 538.

Habeck, D.H., 1974. Caterpillars of Parapoynx in relation to aquatic plants in Florida. Hyacinth Control J 12, 15-18.

Habeck, D.H., 1996. Australian for hydrilla control (Technical Report A-96-10). US Army Corps of Engineers, Waterways Experiment Station, Gainesville, Florida, pp. 40.

Habeck, D.H., Cuda, J.P., 2010. Waterlily Leafcutter, Synclita obliteralis (Walker)(Insecta: Lepidoptera: Crambidae: Acentropinae). Featured Creatures from the Entomology and Nematology Department, Florida Cooperative Extension Service, Institute of Food and Agricultural Sciences, University of Florida. University of Florida, Gainesville, Florida. (http://edis.ifas.ufl.edu/pdffiles/IN/IN80300.pdf).

165

Hall, T.A., 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symposium Series 41, 95- 98.

Hamilton, M.B., 1999. Four primer pairs for the amplification of chloroplast intergenic regions with intraspecific variation. Molecular ecology 8, 521-523.

Hampson, G., 1894. The Fauna of British India, including Ceylon and Burma. Moths. Volume 2. Taylor and Francis, Red Lion Court, Fleet Street.

Handoo, Z.A., Ellington, D., 2005. Nematode Extraction Procedures. Some procedures for collecting and preparing nematodes for study, United Sates Department of Agriculture. Available at http://www.ars.usda.gov/pandp/docs.htm?docid=9942

Harley, K.L.S., Forno, I.W., 1992. Biological control of weeds: a handbook for practitioners and students. Biological control of weeds: a handbook for practitioners and students., 74 pp.

Harris, P., 1973. The selection of effective agents for biological control of weeds. Canadian Entomologist 105, 1495-1503.

Heinrichs, E., Barrion, A.T., 2004. Rice-feeding insects and selected natural enemies in West Africa: biology, ecology, identification. IRRI- the International Rice Research Inststitue, Manilla, Phillippines.

Hijmans, R.J., Cameron, S.E., Parra, J.L., Jones, P.G., Jarvis, A., 2005. Very high resolution interpolated climate surfaces for global land areas. International Journal of Climatology 25, 1965-1978.

Hijmans, R.J., Guarino, L., Cruz, M., Rojas, E., 2001. Computer tools for spatial analysis of plant genetic resources data: 1. DIVA-GIS. Plant Genetic Resources Newsletter, 15-19.

Hill, M., 1998. Life history and laboratory host range of Stenopelmus rufinasus, a natural enemy for Azolla filiculoides in South Africa. Biocontrol 43, 215-224.

Hillsenhoff, W.L., 1973. Notes on Dubiraphia (Coleoptera: Elniidae) with descriptions of five new species. Annals of the Entomological Society of America 66, 55-61.

Hjalten, J., 2004. Simulating herbivory: Problems and possibilities. Insects and Ecosystem Function 173, 243 - 255.

Hoelmer, K.A., Kirk, A.A., 2005. Selecting biological control agents against arthropod pests: Can the science be improved to decrease the risk of releasing ineffective agents? Biological Control 34, 255-264.

Hoffmann, W.A., Poorter, H., 2002. Avoiding bias in calculations of relative growth rate. Annals of Botany 90, 37-42.

166

Hussner, A., Josephs, M., Schmitz, U., 2007. Occurrences of Hygrophila polysperma (Roxb.) T. Anderson, Pontederia cordata in North Rhine-Westphalia. Floristische Rundbriefe 40, 25-30.

Hussner, A., Van De Weyer, K., Gross, E.M., Hilt, S., 2010. Comments on increasing number and abundance of non-indigenous aquatic macrophyte species in Germany. Weed Research 50, 519-526.

Innes, W.T. 1947. Hygrophila, a new aquarium plant. Aquarium 16:30.

Jacono, C.C., 2003. Hygrophila polysperma (Roxb.) T. Anders. Available at http://nas.er.usgs.gov/taxgroup/plants/docs/hy_polys.html. Nonindigenous Aquatic Species, United States Geological Services.

Jakobsson, M., Rosenberg, N.A., 2007. CLUMPP: a cluster matching and permutation program for dealing with label switching and multimodality in analysis of population structure. Bioinformatics 23, 1801.

Jolivet, P., 1998. Interrelationship between insects and plants. Interrelationship between insects and plants., xii + 309 pp.

Jones, J.W., Luten, J.C., 1988. Chap. 2. Simulation of biological processes. In: Peart, M., Robert, M., Curry, B.R., (Eds.), Agricultural systems modeling and simulation. Dekker Inc., New York.

Joyner, T.A., Lukhnova, L., Pazilov, Y., Temiralyeva, G., Hugh-Jones, M.E., Aikimbayev, A., Blackburn, J.K., 2010. Modeling the Potential Distribution of Bacillus anthracis under Multiple Climate Change Scenarios for Kazakhstan. PLoS ONE 5, e9596.

Julien, M., White, G., 1997. Biological control of weeds: theory and practical application. ACIAR Monograph Series 49, i-iv, 1-190.

Julien, M.H., Griffiths, M.W., 1998. Biological control of weeds, Fourth edition: A world catalogue of agents and their target weeds. CABI Publishing, Wallingford, Oxfordshire, UK

Kalinowski, S.T., 2004. Counting alleles with rarefaction: private alleles and hierarchical sampling designs. Conservation Genetics 5, 539-543.

Kalinowski, S.T., 2005. HP-RARE 1.0: a computer program for performing rarefaction on measures of allelic richness. Molecular Ecology Notes 5, 187-189.

Karban, R., 1989. Fine-scale adaptation of herbivorous thrips to individual host plants. Nature 340, 60-61.

167

Kasselmann, C., 1994. Decorative aquarium plants: Hygrophila polysperma (Roxburgh) T. Anderson. The Aquatic Gardener: Journal of Aquatic Gardeners Association 7, 107-108-113.

Keane, R.M., Crawley, M.J., 2002. Exotic plant invasions and the enemy release hypothesis. Trends in Ecology & Evolution 17, 164-170.

Kehimkar, I., 2008. The Book of Indian Butterflies. Bombay Natural History Society and Oxford University Press, UK.

Kejval, Z., 2003. The genus Anthelephila (Coleoptera: Anthicidae). European Journal of Entomology 100, 381-392.

Knevel, I.C., Lans, T., Menting, F.B.J., Hertling, U.M., van der Putten, W.H., 2004. Release from native root herbivores and biotic resistance by soil pathogens in a new habitat both affect the alien Ammophila arenaria in South Africa. Oecologia 141:502-510.

Kok, L., 2001. Classical biological control of nodding and plumeless thistles. Biological Control 21, 206-213.

Kolmer, J.A., Ordonez, M.E., Groth, J.V., 2009. The Rust Fungi. In: Encyclopedia of Life Sciences (ELS). John Wiley & Sons, Ltd. Chichester.

Kovach, C.W., Kurdziel, J.P., Bowman, R., Wagner, J., Lawrence, J.M., 1992. The Effects of Stress and Disturbance on Proximate Composition, Allocation of Production, Photosynthesis, Respiration, and Chlorophyll Levels in Hygrophila polysperma (Roxb) Anders (Acanthaceae). Environmental and Experimental Botany 32, 479-486.

Kriticos, D.J., 2003. The roles of ecological models in evaluating weed biological control agents and projects. In: Jacob, H.S., Briese, D.T. (Eds.), Improving the selection, testing and evaluation of weed biological control agents. Proceedings of the CRC for Australian Weed Management biological control of weeds symposium and workshop. CRC for Australian Weed Management: Technical Series # 7, Perth, Australia, pp. 69 - 74.

Krombholz, P., 1996. Hygrophila polysperma: an indicator plant. The Aquatic Gardener: J. Aquat. Gardeners Assoc 9, 135-137.

Kumari, S., Lyla, K., 2001. A survey of the pests of orchids. Journal of Tropical Agriculture 39, 32-34.

Landsberg, J., 1990. Dieback of rural eucalypts: Does insect herbivory relate to dietary quality of tree foliage? Australian Journal of Ecology 15, 73-87.

Langeland, K.A., Burks, K.C. (Eds.), 1999. Identification and biology of non-native plants in Florida‟s natural areas. University of Florida, Gainesville, Florida.

168

Laundon, G.F., 1963. Rust fungi, I. On acanthaceae. Mycological papers, C.M.I. 89, 1- 89.

Lavergne, S., Molofsky, J., 2007. Increased genetic variation and evolutionary potential drive the success of an invasive grass. Proceedings of the National Academy of Sciences 104, 3883.

Lehtila, K., Boalt, E., 2004. The use and usefulness of artificial herbivory in plant- herbivore studies. Insects and Ecosystem Function 173, 257-275.

Les, D.H., Wunderlin, R.P., 1981. Hygrophila polysperma Acanthaceae in Florida, USA. Florida Scientist 44, 189-192.

Littlefield, J.L., Buckingham, G.R., 2004. Host specificity testing of biological control agents of weeds. In: Coombs, E.M., Clark, J.K., Piper, G.L., Confrancesco, A.F., (Eds.), Biological control of invasive plants in the United States. Oregon State University Press, Corvallis, Oregon.

Loague, K., Green, R.E., 1991. Statistical and graphical methods for evaluating solute transport models: Overview and application. Journal of Contaminant Hydrology 7, 51-73.

Lobo, J.M., Jiménez-Valverde, A., Real, R., 2008. AUC: a misleading measure of the performance of predictive distribution models. Global Ecology and Biogeography 17, 145-151.

Lodge, D.M., 1991. Herbivory on fresh-water macrophytes. Aquatic Botany 41, 195-224.

Lokker, C., Susko, D., Lovett-Doust, L., Lovett-Doust, J., 1994. Population genetic structure of Vallisneria americana, a dioecious clonal macrophyte. American Journal of Botany 81, 1004-1012.

Long, R.W., 1970. Genera of Acanthaceae in Southeastern United-States. Journal of the Arnold Arboretum 51, 257-&.

Lowe, A., Harris, S., Harris, S.E., Ashton, P., 2004. Ecological genetics: design, analysis, and application. Wiley-Blackwell.

Magurran, A.E. 2004. Measuring biological diversity. Blackwell Publishing.

Manrique, V., Cuda, J., Overholt, W., Williams, D., Wheeler, G., 2008. Effect of host- plant genotypes on the performance of three candidate biological control agents of Schinus terebinthifolius in Florida. Biological Control 47, 167-171.

Manrique, V., Cuda, J.P., Overholt, W.A., Diaz, R., 2008. Temperature-dependent development and potential distribution of Episimus utilis (Lepidoptera : Tortricidae), a candidate biological control agent of Brazilian peppertree (Sapindales : Anacardiaceae) in Florida. Environmental Entomology 37, 862-870.

169

Mathew, G., 2006. An inventory of Indian pyrlids (Lepidoptera: Pyralidae). Zoo's Print Journal 21, 2245-2258.

Mayer, D.G., Butler, D.G., 1993. Statistical validation. Ecological Modelling 68, 21-32.

McCauley, D., Smith, R., Lisenby, J., Hsieh, C., 2003. The hierarchical spatial distribution of chloroplast DNA polymorphism across the introduced range of Silene vulgaris. Molecular ecology 12, 3227-3235

McClay, A.S., Balciunas, J.K., 2005. The role of pre-release efficacy assessment in selecting classical biological control agents for weeds - applying the Anna Karenina principle. Biological Control 35, 197-207.

McClay, A.S., Hughes, R.B., 1995. Effects of temperature on developmental rate, distribution, and establishment of Calophasia lunula (Lepidoptera, Noctuidae), a biocontrol agent for Toadflax (Linaria spp). Biological Control 5, 368-377.

McDade, L.A., Moody, M.L., 1999. Phylogenetic relationships among Acanthaceae: evidence from noncoding trnL-trnF chloroplast DNA sequences. American Journal of Botany 86, 70.

McEvoy, P.B., 1996. Host specificity and biological pest control. Bioscience 46, 401- 405.

McEvoy, P.B., Coombs, E.M., 1999. Biological control of plant invaders: regional patterns, field experiments, and structured population models. Ecological Applications 9, 387-401.

McFadyen, R.E.C., 1998. Biological control of weeds. Annual Review of Entomology 43, 369 - 393.

McNyset, K.M., 2005. Use of ecological niche modelling to predict distributions of freshwater fish species in Kansas. Ecology of Freshwater Fish 14, 243 - 255.

McNyset, K.M., Blackburn, J.K., 2006. Does GARP really fail miserably? A response to Stockman et al. (2006). Diversity and Distributions 12, 782-786.

Merritt, R.W., Cummins, K.W. (Eds.), 1996. An introduction to the aquatic insects of North America. Kendall Hunt.

Misra, B.C., 1980. The leaf and plant hoppers of rice. Central Rice Research Institute, , India.

Mogensen, H.L., 1996. The hows and whys of cytoplasmic inheritance in seed plants. American Journal of Botany, 383-404.

170

Mora-Olivo, A., Daniel, T.F., Martinez, M., 2008. First record in the Mexican flora of Hygrophila polysperma (Acanthaceae), an aquatic weed. Revista Mexicana De Biodiversidad 79, 265-269.

Müeller-Schaerer, H., Schaffner, U., 2008. Classical biological control: exploiting enemy escape to manage plant invasions. Biological Invasions 10, 859-874.

Mukherjee, A., Cuda, J.P., Overholt, W.A., Ellison, C., 2008. Biological control of Hygrophila polysperma: searching for natural enemies in India - First trip report. Aquatics 30, 20-21-22.

Murihead, J.R., Gray, D.K., Kelly, D.W., Ellis, S.M., Heath, D.D., Macisaac, H.J., 2008. Identifying the source of species invasions: sampling intensity vs. genetic diversity. Molecular ecology 17, 1020-1035.

Myers, J.H., Bazely, D.R., 2003. Ecology and control of introduced plants. Ecology and control of introduced plants, xiv + 313 pp.

Nakamura, K., Abbas, I., 1987. Preliminary life table of the spotted tortoise beetle, Aspidomorpha miliaris (Coleoptera: Chrysomelidae) in Sumatra. Researches on population ecology 29, 229-236.

O'Brien, C., Askevold, I., 1995. Systematics and evolution of weevils of the genus Bagous Germar (Coleoptera: Curculionidae). V. Taxonomic treatment of the species of the Indian subcontinent. Contributions of the American Entomological Institute (USA) 28, 1-184.

Owens, C.S., Madsen, J.D., Smart, R.M., Stewart, R.M., 2001. Dispersal of native and nonnative aquatic plant species in the San Marcos River, Texas. Journal of Aquatic Plant Management 39, 75-79.

Page, A.R., Lacey, K.L., 2006. Economic impact assessment of Australian weed biological control. Technical Series - CRC for Australian Weed Management, 150 pp.

Peakall, R., Smouse, P., 2006. GENALEX 6.1: genetic analysis in Excel. Population genetic software for teaching and research. Mol Ecol Notes 6, 288-295.

Pearson, D.E., Callaway, R.M., 2006. Biological control agents elevate hantavirus by subsidizing deer mouse populations. Ecology Letters 9, 443-450.

Pearson, R.G., Thuiller, W., Arau´jo, M.B., Martinez-Meyer, E., Brotons, L.S., McClean, C., Miles, L., Segurado, P., Dawson, T.P., Lees, D.C., 2006. Model-based uncertainty in species range prediction. Journal of Biogeography 33, 1704 -1711.

Peterson, A.T., Papes, M., Eaton, M., 2007. Transferability and model evaluation in ecological niche modeling: a comparison of GARP and Maxent. Ecography 30, 550-560.

171

Peterson, A.T., Papes, M., Kluza, D.A., 2003. Predicting the potential invasive distributions of four alien plant species in North America. Weed Science 51, 863- 868.

Peterson, A.T., Shaw, J., 2003. Lutzomyia vectors for cutaneous leishmaniasis in Southern Brazil: ecological niche models, predicted geographic distributions, and climate change effects. International Journal for Parasitology 33, 919-931.

Phillips, S.J., Anderson, R.P., Schapire, R.E., 2006. Maximum entropy modeling of species geographic distributions. Ecological Modelling 190, 231-259.

Phillips, S.J., Dudik, M., 2008. Modeling of species distributions with Maxent: new extensions and a comprehensive evaluation. Ecography 3, 161-175.

Phillips, S.J., Dudik, M., Elith, J., Graham, C.H., Lehmann, A., Leathwick, J., Ferrier, S., 2009. Sample selection bias and presence-only distribution models: implications for background and pseudo-absence data. Ecological Applications 19, 181-197.

Pimentel, D., Lach, L., Zuniga, R., Morrison, D., 2000. Environmental and economic costs of nonindigenous species in the United States. Bioscience 50, 53-65.

Pimentel, D., Zuniga, R., Morrison, D., 2005. Update on the environmental and economic costs associated with alien-invasive species in the United States. Ecological Economics 52, 273-288.

Powell, R., Myers, J.H., 1988. The effect of Sphenoptera jugoslavica Obenb.(Col., Buprestidae) on its host plant Lam.(Compositae). Journal of Applied Entomology 106, 25-45.

Powell, R.D., 1989. The functional forms of density-dependent birth and death rates in diffuse knapweed (Centaurea diffusa) explain why it has not been controlled by Urophora affinis, U. quadrifasciata, and Sphenoptera jugoslavica. In: Delfosse, E.S. (Ed.), Proceedings of the VII International Symposium on Biological Control of Weeds. DSIR/CSIRO, Rome, Italy, pp. 195-202.

Pritchard, J., Wen, X., Falush, D., 2009. Documentation for structure software: Version 2.3. Universtiy of Chicago, Chicago. See http://pritch.bsd.uchicago.edu/structure_software/release_versions/v2.3.3/structure _doc.pdf, 1–37.

Pritchard, J.K., Stephens, M., Donnelly, P., 2000. Inference of population structure using multilocus genotype data. Genetics 155, 945-959.

Putten, W.H., G. Yeates, H. Duyts, C.S. Reis, G. Karssen. 2005. Invasive plants and their escape from root herbivory: a worldwide comparison of the root-feeding nematode communities of the dune grass Ammophila arenaria in natural and introduced ranges. Biological Invasions 7:733-746.

172

Quimby, P., Frick, K., Wauchope, R., Kay, S., 1979. Effects of cadmium on two biocontrol insects and their host weeds. Bulletin of Environmental Contamination and Toxicology 22, 371-378.

Raes, N., Steege, H. T., 2007. A null-model for significance testing of presence-only species distribution models. Ecography. 30, 727-736.

Raghu, S., Dhileepan, K., Trevino, M., 2006a. Response of an invasive liana to simulated herbivory: implications for its biological control. Acta Oecologica- International Journal of Ecology 29, 335-345.

Raghu, S., Wilson, J.R., Dhileepan, K., 2006b. Refining the process of agent selection through understanding plant demography and plant response to herbivory. Australian Journal of Entomology 45, 308-316.

Ramey, V. 1990 Florida prohibited aquatic plants: Florida Department of Natural Resources by authority of Florida state statutes 369.25, Tallahassee, Florida, USA: Florida Department of Natural Resources.

Rao, V.R.S., Chalam, M., 2007. Biodiversity of planthopper fauna (Delphacidae: Hemiptera) associated with rice and sugarcane crop-ecosystems in South India. 14, 129-141.

Rataj, K., Horeman, T.J., 1977. Aquarium plants. T. F. H. Publ., Inc., Jersey City, New Jersey.

Rayamajhi, M.B., Pratt, P.D., Center, T.D., Tipping, P.W., Van, T.K., 2008. Aboveground biomass of an invasive tree melaleuca (Melaleuca quinquenervia) before and after herbivory by adventive and introduced natural enemies: a temporal case study in Florida. Weed Science 56, 451-456.

Reams, W.M., 1953. The Occurrence and Ontogeny of Hydathodes in Hygrophila polysperma T Anders. New Phytologist 52, 8-13.

Robertson, M.P., Kriticos, D.J., Zacharlades, C., 2008. Climate matching techniques to narrow the search for biological control agents. Biological Control 46, 442-452.

Roelfs, A., 1985. Wheat and rye stem rust. In: Roelfs, A., Bushnell, W.R., (Eds.), The Cereal Rust: Disease, Distribution, Epidemiology and Control. Academic Press, New York, pp. 3-37.

Rosenberg, N.A., 2004. DISTRUCT: a program for the graphical display of population structure. Molecular Ecology Notes 4, 137-138.

SAS Institute. 2002. SAS/STAT User‟s guide. SAS Institute, Cary, North Carolina.

Schaffner, U., 2001. Host range testing of insects for biological weed control: how can it be better interpreted? Bioscience 51, 951-959.

173

Schmitz, D.C., 1985. Hygrophila polysperma - a review of the scientific literature. Florida Department of Natural Resources, Bureau of Aquatic Plant Management, Tallahassee, Florida, USA, pp. 1-2-14.

Schmitz, D.C., 1990. The invasion of exotic aquatic and wetland plants in Florida: history and efforts to prevent new introductions. Aquatics 12, 6-7-16.

Schmitz, D.C., Nall, L.E., 1984. Status of Hygrophila polysperma in Florida. Aquatics 6, 11-12-14.

Scott, J., Adair, R., 1990. The commencement of biological control of bitou bush and boneseed (Chrysanthemoides monilifera). In: Heap, J.W. (Ed.), Proceedings of the 9th Australian weed Confereance, Adelaide, South Australia, August 6-10, 1990, pp. 126-129.

Seal, D.R. 2004 Management of melon thrips, Thrips palmi Karny (Thysanoptera: Thripidae): an integrated approach using chemical, cultural, and biological agents, Proceedings of the Florida State Horticultural Society 117: 63-68

Selkoe, K.A., Toonen, R.J., 2006. Microsatellites for ecologists: a practical guide to using and evaluating microsatellite markers. Ecology Letters 9, 615-629.

Senaratne, K., Palmer, W.A., Sutherst, R.W., 2006. Use of CLIMEX modelling to identify prospective areas for exploration to find new biological control agents for prickly acacia. Australian Journal of Entomology. 45, 298-302.

Senaratne, K., Palmer, W.A., Sutherst, R.W., 2006. Use of CLIMEX modelling to identify prospective areas for exploration to find new biological control agents for prickly acacia. Australian Journal of Entomology 45, 298-302.

Shah, A., Purohit, M., Patel, M., Patel, C., 1990. New record of Pachnephorus sp.(: Chrysomelidae) as a pest of sugarcane in Gujarat. Indian Journal of Entomology 52, 159-160.

Shaw, J., Lickey, E.B., Beck, J.T., Farmer, S.B., Liu, W.S., Miller, J., Siripun, K.C., Winder, C.T., Schilling, E.E., Small, R.L., 2005. The tortoise and the hare II: Relative utility of 21 noncoding chloroplast DNA sequences for phylogenetic analysis. American Journal of Botany 92, 142-166.

Simberloff, D., Schmitz, D.C., Brown, T.C., 1997. Strangers in paradise: impact and management of nonindigenous species in Florida. Island Press.

Soti, P.G., Volin, J.C., 2010. Does water hyacinth (Eichhornia crassipes) compensate for simulated defoliation? Implications for effective biocontrol. Biological Control 54, 35 - 40.

174

Spencer, W., Bowes, G., 1985. Limnophila and Hygrophila - A Review and Physiological Assessment of their Weed Potential in Florida. Journal of Aquatic Plant Management 23, 7-16.

Steinbauer, M.J., 2001. Specific leaf weight as an indicator of juvenile leaf toughness in Tasmanian bluegum (Eucalyptus globulus ssp. globulus): implications for insect defoliation. Australian Forestry 64, 32-37.

Stewart, C.A., Emberson, R.M., Syrett, P., Moran, V.C., Hoffmann, J.H., 1996. Temperature effects on the alligator weed flea-beetle, Agasicles hygrophila (Coleoptera: Chrysomelidae): implications for biological control in New Zealand. Proceedings of the 9th International Symposium on Biological Control of Weeds. 19-26 January 1996, Stellenbosch, South Africa., 393-398.

Stockwell, D., Peters, D., 1999. The GARP modelling system: problems and solutions to automated spatial prediction. International Journal of Geographical Information Science. 13, 143-158.

Strauss, S.Y., Agrawal, A.A., 1999. The ecology and evolution of plant tolerance to herbivory. Trends in Ecology & Evolution 14, 179 - 185.

Subramaniam, P., 1920. Life History of the Buprestid Leaf miner (Trachys bicolor, Kerremans), a Pest on Butea frondosa in Mysore. Journal of the Bombay Natural History Society 27, 178-179.

Sun, Y., Ding, J., Rena, M., 2009. Effects of simulated herbivory and resource availability on the invasive plant, Alternanthera philoxeroides in different habitats. Biological Control 48, 287-293.

Sutton, D. L., 1995. Hygrophila is replacing hydrilla in South Florida. Aquatics. 17, 4-6, 8, 10.

Sutton, D.L., 1996. Life cycle and phenology of hygrophila in relation to development of management strategies. In: Stocker, R.K., (Ed.), Control technologies for use against the submersed aquatic weeds hydrilla and hygrophila. Semi-Annual Report, USDA/ARS-IFAS/University of Florida Cooperative Agreement, USDA- ARS No. 58-6629-4-008. Center for Aquatic Plants, Institute of Food and Agricultural Sciences, University of Florida, Gainesville, Florida., pp. 51-68.

Sutton, D.L., Bitting, L.E., Moore, W.H., Baker, G.E., 1994. Summer treatment of Hygrophila with endothall in South Florida. Aquatics 16, 4-6, 8.

Sutton, D.L., Vandiver Jr, V.V., 2006. Grass carp: A fish for biological management of hydrilla and other aquatic weeds in Florida. Bulletin 867, Department of Fisheries and Aquacultural Sciences, Florida Cooperative Extension Service, Institute of Food and Agricultural Sciences, University of Florida. First published December 1986. Reviewed September 2006. http://edis.ifas.ufl.edu/fa043.

175

Sutton, D.L., Vandiver, V.V.J., 1986. Grass carp: a fish for biological management of hydrilla and other aquatic weeds in Florida. Florida Agriculture Experiment Station Bulletin 867.

Swofford, D., 1999. PAUP*: phylogenetic analysis using parsimony (* and other methods), version 4.02 b. Sinauer, Sunderland, Massachusetts, USA.

Taberlet, P., Gielly, L., Pautou, G., Bouvet, J., 1991. Universal primers for amplification of three non-coding regions of chloroplast DNA. Plant Molecular Biology 17, 1105- 1109.

Thirumalachar, M., Narasimhan, M., 1954. Morphology of Spore Forms and Heteroecism in Puccinia cacao. Mycologia 46, 222-228.

Thornley, J.H.M., Johnson, I.R., 1990. Plant and crop modelling: a mathematical approach to plant and crop physiology. Oxford University Press, Oxford, UK.

Thuiller, W., Broennimann, O., Hughesw, G., Alkemade, J.R.M., Midgley, G.F., Corsi, F., 2006. Vulnerability of African mammals to anthropogenic climate change under conservative land transformation assumptions. Global Change Biology 12, 424- 440.

Tinoco, B.A., Astudillo, P.X., Latta, S.C., Graham, C.H., 2009. Distribution, ecology and conservation of an endangered Andean hummingbird: the Violet-throated Metaltail (Metallura baroni). Bird Conservation International 19, 63-76.

Treadwell, L.W., Cuda, J.P., 2007. Effects of defoliation on growth and reproduction of Brazilian peppertree (Schinus terebinthifolius). Weed Science 55, 137-142.

Trethowan, P. D., Robertson, M.P., McConnachie, A.J., 2010. Ecological niche modelling of an invasive alien plant and its potential biological control agents. South African Journal of Botany. In Press, Corrected Proof.

USDA, 2006 Federal Noxious Weed List, United States Department of Agriculture. Available at http://plants.usda.gov/java/noxious?rptType=State&statefips=12

USDA, 2010. Germplasm Resources Information Network - (GRIN), USDA, ARS, National Genetic Resources Program. National Germplasm Resources Laboratory, Beltsville, Maryland.

Vandiver, V. V., Jr., 1980. Hygrophila. Aquatics. 2, 4-5-11.

Veenakumari, K., Mohanraj, P., Bandyopadhyay, A., 1997. Insect herbivores and their natural enemies in the mangals of the Andaman and Nicobar islands. Journal of Natural History 31, 1105-1126.

Veloz, S.D., 2009. Spatially autocorrelated sampling falsely inflates measures of accuracy for presence-only niche models. Journal of Biogeography 36, 2290-2299.

176

Volkovitsh, M., Dolgovskaya, M.Y., Reznik, S.Y., Markin, G., Cristofaro, M., Tronci, C., 2009. Sphenoptera foveola (Buprestidae) as a potential agent for biological control of skeletonweed, Chondrilla juncea. CABI Publishing, pp. 227.

Wallach, D., 2006. Chapter 2: Evaluating crop models. In: Wallach, D., Makowski, D., Jones, J.W. (Eds.), Working with dynamic crop models: Evaluation, analysis, parameterization, and application. Elsevier, Oxford, UK, pp. 11 - 53.

Wapshere, A. J., 1993. Climate and the prospects for biological-control of weeds - contrasting examples of agents for Solanum elaeagnifolium and Heliotropium amplexicaule. Pest Control & Sustainable Agriculture. 200-202.

Wapshere, A., 1990. Discussion of the biological control of crowsfoot grass,(Eleusine indica), in India, Africa and Australia. In: Heap, J.W. (Ed.), Proceedings of the 9th Australian Weeds Conference, Adelaide, South Austalia, August 6-10, 1990, pp. 484-489.

Wapshere, A.J., 1983. Discovery and testing of a climatically adapted strain of Longitarsus jacobaeae [Col, Chrysomelidae] for Australia. Entomophaga 28, 27- 32.

Ward, S., Gaskin, J., Wilson, L., 2008. Ecological genetics of plant invasion: what do we know? Invasive Plant Science and Management 1, 98-109.

Wheeler, G.S., Center, T.D., 2007. Hydrilla stems and tubers as hosts for three Bagous species: Two introduced biological control agents (Bagous hydrillae and B. affinis) and one native species (B. restrictus). Environmental Entomology 36, 409-415.

Williams, D.A., Overholt, W.A., Cuda, J.P., Hughes, C.R., 2005. Chloroplast and microsatellite DNA diversities reveal the introduction history of Brazilian peppertree (Schinus terebinthifolius) in Florida. Molecular ecology 14, 3643-3656.

Williams, J.N., Seo, C.W., Thorne, J., Nelson, J.K., Erwin, S., O'Brien, J.M., Schwartz, M.W., 2009. Using species distribution models to predict new occurrences for rare plants. Diversity and Distributions 15, 565-576.

Williams, J.R., 1954. The biological control of weeds. Rep. 6th Commonw. Ent. Conf., pp. 95-101.

Willmott, C.J., 1981. On the validation of models. Physical Geography 2, 184 - 194.

Wilson, F., 1964. Biological control of weeds. Annual Review of Entomology 9, 225-&.

Wilson, J.R., Rees, M., Holst, N., Thomas, M.B., Hill, G., 2000. Water hyacinth population dynamics. Biological and integrated control of water hyacinth: Eichhornia crassipes. Proceedings of the Second Meeting of the Global Working Group for the Biological and Integrated Control of Water Hyacinth, 96-104.

177

Wisler, G.C., R.F. Norris. 2005. Interactions between weeds and cultivated plants as related to management of plant pathogens. Weed Science 53:914-917.

Wolda, H. 1981. Similarity indices, sample size and diversity. Oecologia 50:296-302.

Woolfe, T., 1995. Water weed is latest menace. Tallahassee Democrat.

Wright, A., Boland, N., Kerr, J., 1989. Changes in leaf hardness and moisture content of water hyacinth, Eichhornia crassipes, following feeding by Neochetina eichhorniae Warner (Coleoptera: Curculionidae). Bulletin of Entomological Research 79, 529- 535.

Wunderlin, R.P., Hansen, B.F., 2003. Guide to the vascular plants of Florida. University Press of Florida.

Yost, A.C., Petersen, S.L., Gregg, M., Miller, R., 2008. Predictive modeling and mapping sage grouse (Centrocercus urophasianus) nesting habitat using Maximum Entropy and a long-term dataset from Southern Oregon. Ecological Informatics 3, 375-386.

178

BIOGRAPHICAL SKETCH

Abhishek Mukherjee was born in November, 1977, in Narendrapur, West Bengal,

India. He has earned his Bachelor of Science in agriculture in 2001 from the Uttar

Banga Krishi Viswavidyalya, West Bengal India with specialization in agricultural entomology. In 2003, Abhishek earned his Master of Science in agricultural entomology from the Bidhan Chandra Krishi Viswavidyalaya, West Bengal, India. In both his bachelor‟s and master‟s degrees, he was selected to receive the University Merit

Scholarship to cover his tuition. After completion of MS degrees, he worked in several institutions, including as a Research Scholar in a project involving insecticide resistance management of cotton bollworm in the Ramakrishna Ashram Krishi Vigyan Kendra,

West Bengal and as an Agricultural Officer in the Bank of Baroda, Uttar Pradesh, India.

In 2007, he joined the Entomology and Nematology Department, University of Florida as a PhD student under Dr. James P. Cuda. In his PhD dissertation research he studied the prospects for classical biological control of the invasive aquatic weed hygrophila. As part of this project he has conducted extensive field surveys in India and Bangladesh, native habitats of hygrophila, to identify candidate natural enemies. He has participated in numerous domestic and international conferences (eleven). Besides his dissertation research he also was involved in a number of collaborative research projects. He has received several travel (four) and research (four) grants. In addition, he regularly participated in Teacher Training Workshops organized by the Center of Aquatic and

Invasive Plants, UF. After completion of his PhD, he will be joining as a postdoctoral research scientist in the Texas A & M University. His long term goal is to stay in academics and conduct research on biological control and other biorational methods of pest management.

179