Habitat Characteristics of Taro Loi and Managed

HABITAT CHARACTERISTICS OF REFUGE WETLANDS AND TARO LO’I

USED BY ENDANGERED WATERBIRDS AT HANALEI NATIONAL

WILDLIFE REFUGE, HAWAI’I

HUGO K. W. GEE

A thesis submitted in partial fulfillment of the requirements for the

Master of Science

Major in Wildlife Sciences

South Dakota State University

2007

HABITAT CHARACTERISTICS OF REFUGE WETLANDS AND TARO LO’I

USED BY ENDANGERED WATERBIRDS AT HANALEI NATIONAL

WILDLIFE REFUGE, HAWAI’I

This thesis is approved as a creditable and independent investigation by a candidate for the Master of Science degree and is acceptable for meeting the thesis requirements for this degree. Acceptance of this thesis does not imply that the conclusions reached by the candidate are necessarily the conclusions of the major department.

______Leigh H. Fredrickson Thesis Advisor Date

______Charles Scalet Head, Wildlife and Fisheries Date

iii

ACKNOWLEDGMENTS

I would like to thank Dr. Leigh H. Fredrickson for providing me with guidance

and opportunities during my education. I especially enjoyed talking with him about

conservation issues and sampling the local cuisine on our travels together. I want to

thank Dr. W. Carter Johnson, Dr. Daniel E. Hubbard, and Dr. Dennis Todey for being on my committee. I also am grateful to Fred Paveglio, Kevin Kilbride, Mike Hawkes,

Brenda Zaun, and Michael Mitchell for their suggestions throughout this study.

Thank you to the U.S. Fish and Wildlife Service and Ducks Unlimited for providing funding to study tropical wetlands, taro lo’i, and endangered waterbirds on

Kaua’i. I appreciate the administrative support provided by the Department of Wildlife and Fisheries Sciences at South Dakota State University. I also would like to thank the staff of Kilauea National Wildlife Refuge Complex for securing housing and transportation for this project.

I am indebted to Ashley B. Hitt for her countless hours of field work and companionship. I acknowledge the efforts of Tandi Perkins and Scott Becker in pilot studies that were critical to this project. I am grateful to Chadd Smith for sharing his knowledge and enthusiasm of wetland management at Hanalei National Wildlife Refuge.

I value Steve Wall’s help with ArcView and Dr. Gary Larson’s assistance with identifying plants. I also appreciate the cooperation and friendship of taro farmers at the refuge during this study.

My time in South Dakota was made memorable by the company of friends such as Sharon N. Kahara, Bernard M. Hien, Sheila K. Thomson, Madeline R. Schickel, Nick

L. Wirwa, Jennifer L. Gutscher, Jessica F. Lee, Kari A. Ranallo, and Karen E. Arnold.

My siblings Maggie, Albert, and Trudy (you too Karen) made trips back to Montreal special by keeping the home fires burning. I am blessed to have two loving parents who tolerate my wandering nature. Special thanks to Anil K. Patel, Ming Cheung, Peter

Levidis, and Ngaio Richards for their moral support throughout the years. I would especially like to thank Grandma Gee for always pushing me to new limits. Finally, this thesis is dedicated to Grandma and Grandpa Tam who will always be in my thoughts.

ABSTRACT

HABITAT CHARACTERISTICS OF REFUGE WETLANDS AND TARO LO’I

USED BY ENDANGERED WATERBIRDS AT HANALEI NATIONAL

WILDLIFE REFUGE, HAWAI’I

HUGO K. W. GEE

2007

Hanalei National Wildlife Refuge (HNWR) was established to protect habitat for

the endangered Hawaiian common moorhen (Gallinula chloropus sandvicensis), Hawaiian coot (Fulica alai), Hawaiian duck or Koloa (Anas wyvilliana), and Hawaiian stilt

(Himantopus mexicanus knudseni). I studied two major habitat types (refuge wetlands and

taro lo’i) used by these endangered waterbirds (EWBs) as well as wetland vegetation and

EWB response to moist-soil management at HNWR. Furthermore, the links are discussed

between habitat conditions that occur during the taro agricultural cycle and how each stage

contributed to the life-history requirements for foraging, loafing, and nesting of these four

EWBs.

During the study, effective management of refuge wetlands at HNWR was not possible because the water control infrastructure did not allow for the effective transfer or discharge of water needed to promote optimal habitat conditions for all EWBs. As a result, invasive plant species dominated most of the wetland management units. A few wetland subunits (3.3 ha) did have rototiller treatments combined with water level manipulations in an attempt to create conditions for the germination of native and/or naturalized vegetation.

These manipulations produced moist-soil vegetation such as the annual sedge, fimbry

(Fimbristylis littoralis), or the perennial knotgrass (Paspalum distichum). Moorhen were commonly observed foraging in F. littoralis or P. distichum. Koloa also were observed feeding and loafing in F. littoralis after water levels were raised to overtop seedheads.

The majority of EWBs were observed foraging or loafing on grass-covered dikes surrounding taro because frequent mowing encouraged vigorous plant growth and increased visibility of predators. Moorhen preferred lo’i being harvested and stilt preferred unvegetated wet fallow lo’i; whereas, coots and Koloa preferred vegetated wet fallow lo’i. EWBs may prefer open water conditions in these taro habitat categories because they provide important invertebrate food resources for breeding and brood- rearing. Moorhen preferred less intensively managed taro lo’i in early growth or mature and medium to dense growth stages because these habitats provided structural support for overwater nests and the complex habitat structure for invertebrates. Furthermore, annual wetland plants also provide forage for moorhen, coot and Koloa.

Call response surveys indicated that moorhen were more abundant in taro lo’i (3.6 birds/ha) compared to refuge wetlands (1.6 birds/ha); however, wetland infrastructure was not in place throughout the study to promote optimal conditions in all refuge wetlands. Taro lo’i in early growth or mature and medium to dense growth stages provided important cover and obscurity for 41% (24 of 58) of the moorhen nests found throughout the study. Nest success for moorhens was 64%; however, recruitment may have been as low as 2.5%, likely as result of predation.

vii

Wetland management (rototiller treatments and partial water drawdowns) for EWB nesting was implemented on 20.2 ha of refuge wetlands either by partial water drawdowns on subunits (16.9 ha) with limited vegetation cover or by removing dense cover (3.3 ha) with rototiller treatments. These manipulations were important in creating suitable nesting habitat for stilts (42 of 48 nests). Partial water drawdowns within managed wetlands with limited residual vegetation created suitable habitat for nesting stilts in 15 days; whereas, response to rototiller treatments took 42 days. Nest success for stilts was 43% and recruitment was 3%. Predator removal, flood control by refuge staff and taro farmers, and limiting human disturbance can potentially improve nest success rate and EWB recruitment at HNWR.

In conclusion, the availability of taro habitat conditions suitable for EWBs is influenced by the stage of taro that varies with market demand throughout the year. In contrast, refuge staff can manipulate wetland conditions to create EWB feeding or nesting habitat as needed. However, refuge wetlands must be reconfigured into larger units based on soil texture and a water distribution system developed to allow management activities that will result in habitat conditions for all EWB life-history stages.

viii

TABLE OF CONTENTS

Page

ACKNOWLEDGMENTS ...... III

ABSTRACT...... V

LIST OF TABLES...... XIII

LIST OF FIGURES ...... XV

LIST OF APPENDICES...... XVII

INTRODUCTION ...... 1

LITERATURE REVIEW ...... 5

MOIST-SOIL MANAGEMENT AT HANALEI NATIONAL WILDLIFE REFUGE ...... 5

TARO CULTIVATION AT HANALEI NATIONAL WILDLIFE REFUGE...... 7

HAWAIIAN COMMON MOORHEN ECOLOGY ...... 9

HAWAIIAN COOT ECOLOGY ...... 11

HAWAIIAN DUCK (KOLOA) ECOLOGY...... 13

HAWAIIAN STILT ECOLOGY ...... 14

RATIONALE OF STUDY...... 17

STUDY SITE...... 19

STUDY PLOTS ...... 20

Taro lo’i ...... 21

Refuge wetlands...... 24

METHODS ...... 34

MACROHABITAT CHARACTERIZATION ...... 34

Refuge wetlands...... 34

Taro lo’i ...... 34

WATERBIRD MONITORING...... 36

Visual scan surveys...... 36

Visibility index...... 37

Call response surveys...... 41

MICROHABITAT CHARACTERIZATION...... 42

Refuge wetlands...... 42

Taro lo’i ...... 43

NEST ECOLOGY ...... 44

DATA ANALYSIS...... 47

Macrohabitat characterization...... 47

Waterbird surveys ...... 47

Microhabitat characterization ...... 50

Nesting ecology ...... 50

CONSTRAINTS OF STUDY ...... 51

VEGETATION AND ENDANGERED WATERBIRD RESPONSE TO ROTOTILLER TREATMENTS ...... 53

RESULTS ...... 53

Vegetation response to rototiller treatments ...... 53

Endangered waterbird response to rototiller treatments ...... 59

DISCUSSION ...... 62

Vegetation response to wetland management...... 62

Endangered waterbird response to wetland management...... 63

ENDANGERED WATERBIRD RESPONSE TO SEASONAL CHANGES IN TARO HABITAT AVAILABILITY...... 67 RESULTS ...... 67

Taro habitat availability ...... 67

Waterbird use of taro lo’i...... 69

Daily and seasonal patterns of abundance ...... 69

Endangered waterbird use of taro habitat ...... 70

Behavior of endangered waterbirds in taro habitat...... 71

Distribution of waterbirds in relation to water depth in taro lo’i...... 71

Endangered waterbird densities in eight taro habitat categories...... 72

Endangered waterbird preference of eight taro habitat categories...... 72

Endangered waterbird habitat selection by taro farming intensity ...... 73

Vegetation characteristics of taro habitat...... 76

Visibility index...... 77

DISCUSSION ...... 78

Taro habitat availability ...... 78

Waterbird use of taro lo’i...... 78

Daily and seasonal patterns of abundance ...... 78

Endangered waterbird use of taro habitat ...... 80

Distribution of waterbirds in relation to water depth in taro lo’i...... 81

Endangered waterbird taro habitat selection...... 81

ESTIMATING HAWAIIAN COMMON MOORHEN DENSITIES USING CALL- RESPONSE SURVEYS ...... 84 RESULTS ...... 84

DISCUSSION ...... 86

ENDANGERED WATERBIRD NESTING ECOLOGY ...... 89

RESULTS ...... 89

Hawaiian common moorhen...... 89

Hawaiian coot ...... 95

Hawaiian duck (Koloa)...... 96

Hawaiian stilt ...... 97

Endangered waterbird mortality ...... 100

DISCUSSION ...... 101

Hawaiian common moorhen...... 101

Hawaiian coot ...... 103

Hawaiian duck (Koloa)...... 103

Hawaiian stilt ...... 104

Endangered waterbird recruitment...... 105

MANAGEMENT RECOMMENDATIONS AND RESEARCH NEEDS...... 108

xii

MANAGEMENT RECOMMENDATIONS ...... 108

Taro cultivation...... 108

Wetland management...... 110

Dike management ...... 112

Endangered waterbird recruitment...... 113

RESEARCH NEEDS ...... 115

SUMMARY...... 117

LITERATURE CITED ...... 119

xiii

LIST OF TABLES

Page

Table 1: Timing of rototiller and water management treatments in managed wetlands at Hanalei National Wildlife Refuge during 2004...... 6

Table 2: Taro cultivation practices during the taro agricultural cycle at Hanalei NWR...... 8

Table 3: Area farmed for taro by special use permit holders at Hanalei NWR during 2004...... 21

Table 4: Area of refuge wetland impoundments (not including ditches and dikes) at Hanalei NWR during the 2004 field season...... 30

Table 5: Percent of sample plots within taro lo’i managed by taro permitees that had non-taro emergent vegetation based on 2003 data...... 35

Table 6: Modified Daubenmire cover classification (Daubenmire 1959)...... 44

Table 7: Vegetation response after soil disturbance in refuge wetlands at Hanalei NWR during the 2004 field season...... 54

Table 8: Average endangered waterbird density (birds/ha ± SD) for 28 weekly surveys by taro habitat category at Hanalei NWR from 3 March to 12 November 2004...... 73

Table 9: Average difference in ranks and preference of taro habitat categories by usage and availability for that category for Hawaiian common moorhens, coots, ducks, and stilts at Hanalei NWR from 3 March to 12 November 2004...... 74

Table 10: Average difference in ranks and preference of taro habitat categories by usage and availability for that category by farming intensity for Hawaiian common moorhens, coots, ducks, and stilts at Hanalei NWR from 3 March to 12 November 2004 (* = significant difference between farming intensities; p ≤ 0.05)...... 75

Table 11: Hawaiian duck and stilt visibility index (number in survey/number in walk-by) by taro habitat category at Hanalei NWR from 3 March to 12 November 2004...... 77

xiv

Table 12: Hawaiian common moorhen call-response and visual survey results (subtotal/% of total) by habitat type at Hanalei NWR during the 2004 peak and off-peak breeding season...... 85

Table 13: Pooled visual obscurity, percent green cover, and number of plant species (mean ± SD; n) at nest microhabitat sites by taro habitat category of Hawaiian common moorhen nest sites at Hanalei NWR from 2 February to 18 November 2004...... 93

Table 15: Hawaiian common moorhen, Hawaiian duck, and Hawaiian stilt egg and nest metrics (mean ± SD) at Hanalei NWR from 23 January to 18 November 2004...... 94

Table 16: Hawaiian common moorhen and Hawaiian stilt nest predators at Hanalei NWR from 2 February to 18 November 2004...... 94

Table 17: Reasons for Hawaiian common moorhen nest failure in intensively and less intensively managed taro lo’i at Hanalei NWR from 2 February to 18 November 2004...... 95

Table 19: Percent green and litter cover (% mean ± SD) measured at Hawaiian duck (n = 7) and Hawaiian stilt (n = 46) nest microhabitat sites at Hanalei NWR from 23 January to 23 June 2004...... 97

Table 21: Causes of EWB mortalities at Hanalei NWR from 22 January to 2 August 2004...... 101

LIST OF FIGURES

Page

Figure 1: Hawaiian common moorhen or ‘Alae ‘ula (Gallinula chloropus sandvicensis)...... 10

Figure 2: Hawaiian coot or ‘Alae ke’oke’o (Fulica alai)...... 12

Figure 3: Hawaiian duck or Koloa (Anas wyvilliana)...... 13

Figure 4: Hawaiian stilt or Ae’o (Himantopus mexicanus knudseni)...... 15

Figure 5: Landsat 7 image of Hanalei NWR on January 2003...... 20

Figure 6: Distribution of taro lo’i at Hanalei NWR...... 23

Figure 7: Recently planted (<25% taro cover)...... 25

Figure 8: Early growth or mature (25%-50% taro cover)...... 25

Figure 9: Medium to dense growth (>50% taro cover)...... 26

Figure 10: Lo’i being harvested...... 26

Figure 11: Unvegetated wet fallow (≥50% water cover and <25% emergent vegetation cover)...... 27

Figure 12: Vegetated wet fallow (≥50% water cover and ≥25% emergent vegetation cover)...... 27

Figure 13: Unvegetated dry fallow (<50% water cover and <25% emergent vegetation cover)...... 28

Figure 14: Vegetated dry fallow (<50% water cover and ≥25% emergent vegetation cover)...... 28

Figure 15: Infrastructure of refuge wetland units “A”, “B”, and “C” at Hanalei NWR... 29

Figure 16: Soils of refuge wetland units “A”, “B”, and “C” at Hanalei NWR (unpublished data)...... 32

xvi

Figure 17: Scan survey observation sites used at Hanalei NWR from 12 January to 12 November 2004...... 38

Figure 18: View of Fitzgerald taro lo’i (in red) from observation site #2 at Hanalei NWR from 12 January to 12 November 2004...... 39

Figure 19: View of Fitzgerald taro lo’i (in red) from observation site #3 at Hanalei NWR from 12 January to 12 November 2004...... 39

Figure 20: Mowed dike surrounding taro lo’i managed by Haraguchi at Hanalei NWR...... 40

Figure 21: Unmowed dike (in red) between refuge wetlands at Hanalei NWR ...... 40

Figure 22: Average number of species per vegetated sample plot in refuge wetlands “A1”, “B1”, “B4”, and “B6” following soil disturbance and initial flooding at Hanalei NWR during the 2004 field season...... 55

Figure 23: Vegetation cover (% of subunit) of Fimbristylis littoralis and Paspalum distichum in wetland subunits “A1”, “B1”, “B4”, and “B6” following soil disturbance and initial flooding at Hanalei NWR during the 2004 field season...... 57

Figure 24: Average visual obscurity (dm) of vegetation sampled in wetland subunits “A1”, “B1”, “B4”, and “B6” following soil disturbance and initial flooding at Hanalei NWR during the 2004 field season...... 58

Figure 25: Average percent cover within a sampling frame (at surface of water, 15 cm height, and canopy level) of Fimbristylis littoralis on fine and coarse-textured soils following soil disturbance and initial flooding at Hanalei NWR during the 2004 field season...... 60

Figure 26: Average percent cover within a sampling frame (at surface of water, 15 cm height, and canopy level) of Paspalum distichum on fine and coarse-textured soils following soil disturbance and initial flooding at Hanalei NWR during the 2004 field season...... 61

Figure 27: Average area (ha) in taro habitat categories during morning and afternoon surveys at Hanalei NWR from 3 March to 12 November 2004...... 68

Figure 28: Endangered waterbird average numbers during morning and afternoon surveys at Hanalei NWR from 12 January to 12 November 2004...... 70

Figure 29: Number of Hawaiian common moorhen nests by taro habitat category at Hanalei NWR from 2 February to 18 November 2004 (n = 53)...... 92

xvii

LIST OF APPENDICES

Appendix Page

1 Behavior, habitat, and water depth variables recorded during visual scan surveys...... 128

2 Daily precipitation recorded at Princeville Ranch, Hawai’i from 1 January to 30 November 2004 (National Climatic Data Center)...... 129

3 Plants found at refuge wetland microhabitat sites at Hanalei NWR from 23 June to 17 November 2004 (www.plants.usda/gov/)...... 133

4 Habitat type of Hawaiian common moorhen, coot, duck, and stilt observations (number observations/ % of observations) in refuge wetlands (“A1”, “B1”, “B4”, and “B6”) following rototiller treatments in 2004 at Hanalei NWR...... 134

5 Behavior of Hawaiian common moorhen, coot, duck, and stilt observations (number observations/ % of observations) in refuge wetlands (“A1”, “B1”, “B4”, and “B6”) following rototiller treatments in 2004 at Hanalei NWR...... 135

6 Water depth at site of detection for Hawaiian common moorhen, coot, duck, and stilt observations (number observations/ % of observations) in refuge wetlands (“A1”, “B1”, “B4”, and “B6”) following rototiller treatments in 2004 at Hanalei NWR...... 136

7 Area of taro (ha) in 8 habitat categories at Hanalei NWR from 3 March to 12 November 2004...... 137

8 Endangered waterbird observations (total number of observations/%) by taro habitat category at Hanalei NWR from 3 March to 12 November 2004...... 138

9 Hawaiian common moorhen behavior observed (number observations/ % of observations) in taro dikes and habitat categories at Hanalei NWR from 3 March to 12 November 2004...... 139

10 Hawaiian coot behavior observed (number observations/ % of observations) in taro dikes and habitat categories at Hanalei NWR from 3 March to 12 November 2004...... 140

xviii

11 Hawaiian duck behavior observed (number observations/ % of observations) in taro dikes and habitat categories at Hanalei NWR from 3 March to 12 November 2004...... 141

12 Hawaiian stilt behavior observed (number observations/ % of observations) in taro dikes and habitat categories at Hanalei NWR from 3 March to 12 November 2004...... 142

13 Water depth at site of detection for Hawaiian common moorhen observations (number observations/ % of observations) in taro habitat categories at Hanalei NWR from 3 March to 12 November 2004...... 143

14 Water depth at site of detection for Hawaiian coot observations (number observations/ % of observations) in taro habitat categories at Hanalei NWR from 3 March to 12 November 2004 at Hanalei NWR...... 144

15 Water depth at site of detection for Hawaiian duck observations (number observations/ % of observations) in taro habitat categories at Hanalei NWR from 3 March to 12 November 2004...... 145

16 Water depth at site of detection for Hawaiian stilt observations (number observations/ % of observations) in taro habitat categories at Hanalei NWR from 3 March to 12 November 2004...... 146

17 Plants found at EWB high density sites in taro lo’i at Hanalei NWR from 3 March to 12 November 2004 (www.plants.usda/gov/)...... 147

18 Number of plant species (mean ± SD; n) by taro habitat category measured in vegetated sample plots of taro lo’i with the highest densities of Hawaiian common moorhens, coots, ducks, and stilts at Hanalei NWR from 3 March to 12 November 2004...... 148

19 Pooled visual obscurity (mean ± SD; n) by taro habitat category measured in vegetated sample plots of taro lo’i with the highest densities of Hawaiian common moorhens, coots, ducks, and stilts at Hanalei NWR from 3 March to 12 November 2004...... 149

20 Percent green cover (mean ± SD; range; n) of vegetated sample plots by taro habitat category with the highest density of Hawaiian ducks at Hanalei NWR from 3 March to 12 November 2004...... 150

21 Percent green cover (mean ± SD; range; n) of vegetated sample plots by taro habitat category with the highest density of Hawaiian stilts at Hanalei NWR from 3 March to 12 November 2004...... 151

xix

22 Percent green cover (mean ± SD; n) of vegetated sample plots by taro habitat category with the highest density of Hawaiian common moorhens at Hanalei NWR from 3 March to 12 November 2004...... 152

23 Percent green cover (mean ± SD; n) of vegetated sample plots by taro habitat category with the highest density of Hawaiian coots at Hanalei NWR from 3 March to 12 November 2004...... 153

24 List of waterbird species observed during scan surveys at Hanalei NWR during the 2004 field season...... 154

INTRODUCTION

The main Hawaiian Islands have porous volcanic soils and a relative lack of

expansive floodplains which limits the amount of wetland habitat available to endangered

waterbirds (EWBs; Shallenberger 1977). In fact, wetlands comprise only 3% or 21,000

ha of the state’s landmass (U.S. E.P.A. 2003). Nonetheless, the Hawaiian Islands have a

great variety of natural and man-made aquatic habitats. Natural types include upland

bogs, swamps, streams and tributaries, large rivers, inland freshwater marshes, and

coastal brackish marshes (Shallenberger 1977). Among its man-made or manipulated

habitats, there are taro lo’i, managed wetlands, drainage and irrigation ditches, reservoirs and other water storage areas, sewage ponds, cane waste settling ponds, coastal and inland fishponds and other aquaculture facilities as well as ephemerally flooded pastures.

Prior to the arrival of Europeans during the late 18th century, Hawaiians created

additional wetland habitat for waterbirds through the construction of fishponds in coastal

areas and the diversion of streams for the shallow flooding of taro crops referred to as lo’i

(Swedberg 1967, Shallenberger 1977). In fact, the Hanalei Valley on Kaua’i has a history of taro cultivation that stretches back more than 1,300 years (Schilt 1980).

However, an accelerating influx of Asian immigrants led to a shift in diet as taro lo’i was replaced with rice paddies (Coleman 1979). By the early 20th century, Hawaiian rice

cultivation declined as production shifted to California growers. The subsequent

abandonment of these man-made agricultural wetlands and the draining of natural

wetlands for housing and industry eliminated much of the prime habitat of native

waterbirds. The expansion of other forms of agriculture (e.g., sugar cane and pineapple)

also eliminated EWB habitat although it created new habitat in irrigation reservoirs and

cane waste settling basins (Shallenberger 1977). Because the conditions of the artificial waterbird habitats are subject to the demands for water by these industries, these areas are managed without consideration for EWB habitat requirements.

By the 1980s, Hawai’i had lost 12% of its wetland area (Mitsch and Gosselink

1993), where many of the remaining wetlands were degraded. Invasive, exotic plants

(e.g., mangrove and various weedy grasses) have encroached on fishponds and wetlands, which leaves only a fraction of the native habitat in good condition (Scott 1993).

Destruction and alteration of habitat, hunting, introduced predatory mammals and nonnative birds, and diseases have caused the extinction of Hawai’i’s endemic rails,

flightless geese, and an ibis species (Olson and James 1991).

The Hawaiian common moorhen or ‘Alae ‘ula (Gallinula chloropus

sandvicensis), Hawaiian coot or ‘Alae ke’oke’o (Fulica alai), Hawaiian duck or Koloa

(Anas wyvilliana), Hawaiian stilt or Ae’o (Himantopus mexicanus knudseni), Laysan

duck (Anas laysanensis), and Hawaiian goose or Nēnē (Branta sandvicensis) are

Hawai’i’s remaining endemic waterbird species, which are federally listed as endangered.

Recovery actions for the Hawaiian common moorhen, Hawaiian coot, Hawaiian duck, and Hawaiian stilt are described under the Hawaiian Waterbirds Recovery Plan (USFWS

2005). The ultimate goal of the recovery plan is to restore and maintain multiple self- sustaining populations of these 4 EWBs within their historical ranges, which will likely

allow them to be delisted to threatened and eventually removed from the Federal List of

Endangered and Threatened Wildlife and Plants (USFWS 2005).

Under the Endangered Species Act of 1973, the U.S. Fish and Wildlife Service

(USFWS) has a legal responsibility to protect and manage habitat crucial to promote the

recovery of Hawai’i’s EWBs. Because waterbirds regularly use the taro lo’i of the

Hanalei Valley, state and federal biologists recognized the possibility that taro farming

and wetland management were complementary for EWBs (Shallenberger 1977). In 1972,

the USFWS established Hanalei National Wildlife Refuge (HNWR) to protect important

habitat for Hawai’i’s EWBs.

The Hawaiian Waterbirds Recovery Plan identified HNWR as a core wetland that

provides habitat essential to larger populations of Hawaiian EWBs (USFWS 2005).

Management practices at the refuge should include the following in accordance with this

recovery plan: 1) develop a habitat management plan (HMP); 2) implement the HMP by

manipulating water substrates, and vegetation (native and naturalized/exotic) to maximize

nesting success, brood survival, food availability, and recruitment of EWBs; 3) monitor,

eliminate, and/or reduce predator populations such as feral cats (Felis catus), feral dogs

(Canis familiaris), rats (Rattus spp.), cattle egrets (Bulbulcus ibis), and bullfrogs (Rana

catesbeiana); 4) minimize human disturbance to EWBs and their habitats by controlling

human access to waterbird habitats during the breeding season and resolving conflicts

from actual or perceived depredation on agriculture crops by EWBs; 5) monitor EWB

populations to detect disease outbreaks as soon as possible and take immediate action to

restrict the spread of disease; 6) monitor all populations of EWBs with biannual statewide

surveys and regular standardized surveys while developing improved survey techniques

for the Hawaiian duck and common moorhen; 7) monitor EWB reproductive success and

aquatic invertebrate prey species used by EWBs; 8) determine whether fish compete with

EWBs for aquatic invertebrates; and 9) remove the threat of mallard-Hawaiian duck hybridization.

LITERATURE REVIEW

Moist-soil management at Hanalei National Wildlife Refuge

Moist-soil management involves the use of water level control and/or soil

disturbance (mowing, plowing, and discing) to enhance the growth of desired, native

vegetative communities (Fredrickson 1996). Refuge wetlands with dense emergent or

open water habitats were treated with a rototiller fitted with “s-shaped” blades to provide

habitat with an interspersion of early succession (exposed mudflats) to late succession

habitats to satisfy the life-history needs of EWBs as well as a diverse community of other

wetland-dependent biota. Rototiller treatments were conducted in wetland subunits

“A1”, “B1”, “B4”, and “B6” (3.3 ha) during the spring of 2004 (Table 1). Before the

rototiller treatment, the area was mowed if necessary. Rototilling vegetation with “s-

shaped” blades accelerates decomposition and invertebrate responses while creating

mudflat habitat to promote germination of native wetland plant species and other seed-

producing annuals that provide forage and cover for EWBs (Fredrickson 1996). Water

levels were manipulated during the spring and summer of 2004 to eliminate undesirable

upland plants in rototilled wetlands and to mimic natural drying and flooding events to

enhance the availability of aquatic invertebrates (Fredrickson 1996, Davis and Smith

1998, Anderson and Smith 1999 - Table 1).

Table 1: Timing of rototiller and water management treatments in managed wetlands at Hanalei National Wildlife Refuge during 2004. Wetland Objective Treatment Date subunit “A1” Create mudflats for EWB Deep rototilled in 23-25 March foraging, germination of shallow water then drawn annual plants, and stilt down nesting Deep and then shallow 26-31 March rototilled Eliminate undesirable Flooded 1 April upland plants Expose mudflats for Water draw down 6 April plant germination and to create stilt nesting habitat Create aquatic Flooded 13 July invertebrate habitat “B1” Create conditions to Mowed and flooded 10-20 March decompose existing vegetation Create mudflats for EWB Rototilled with standing 22-24 March foraging, germination of water annual plants, and stilt nesting Expose mudflats for Water draw down 6 April plant germination and to create stilt nesting habitat Create aquatic Flooded 16 June invertebrate habitat “B4” Create mudflats for EWB Rototilled 10 April foraging, germination of annual plants, and stilt nesting Eliminate undesirable Flooded 8 May upland plants “B6” Create mudflats for EWB Rototilled 10 April foraging, germination of annual plants, and stilt nesting Eliminate undesirable Flooded 8 May upland plants

Taro cultivation at Hanalei National Wildlife Refuge

Taro (Colocasia esculenta), a member of the Araceae family, is an agricultural crop grown for its edible corms (underground storage organ) and leaves as well as its traditional ceremonial uses (Wang 1983). Taro farming requires ample amounts of cool, fresh water to enhance taro growth. When planting huli (cuttings of the corm with petiole), ponds are flooded to a 5-7½ cm depth and then increased to a 10-15 cm depth when roots and leaves are established (De la Pena 1983). Shallow flooding of lo’i reduces the incidence of corm and root disease caused from Pythium spp. (De la Pena

1983). This water management practice also minimizes encroachment by non-taro emergent vegetation.

Non-taro emergent vegetation is removed manually and chemically (Roundup™) on dike edges 6 to 8 times a year in some refuge taro fields (Table 2). Non-taro emergent

vegetation is removed manually during the first 4-6 months of taro growth. Farmers do not enter fields after 4-6 months to decrease damage to new taro stalks growing from lateral roots. Roundup™ is applied to fields after they are drained and left dry for 2 to 4 weeks.

Vegetation growth is rapid during the first 6 months eventually leading to 95% canopy coverage (Perkins and Gee, unpublished report). Taro’s rapid growth is aided by several applications of nitrogen fertilizer and lime (calcium carbonate), which neutralizes the acidity of the soil. Before planting huli, dry fields may be disced or rototilled and then granular nitrogen (as little as 64 kg/ha) and lime (up to 992 kg/ha) are tilled into the soil. At 3-4 months in the agricultural cycle, nitrogen fertilizer is applied one final time.

X 12 to 16 11

X Month 5

X 4

X X X 3

X X 2

X X 1

X the taro agricultural cycle at Hanalei NWR. Hanalei cycle at agricultural taro the

X X X X X Before planting zer ili ert f trogen trogen i y n l pp

Table 2: Taro cultivation practices during (2 times) (2 times) Apply Lime (2-3 times) Plowing Rototilling Planting huli Manual/chemical weeding (6-8 times) Harvest practice Taro cultivation A

Farmers drain fields so that nitrogen is not washed away and fields are reflooded after 2-3

days. Farmers also drain fields as they apply lime up to two more times through aerial

dispersal (fired by a fertilizer air blower). During months 7 to 12, leaves diminish in size

and abundance which facilitate harvesting. Corm development begins between months 3

to 5 and continues until harvest.

Taro lo’i are dried for one week to loosen corms and reflooded before being

harvested manually. Harvesting is completed in 1-4 weeks depending on the size of the farmer’s lo’i and the number of helpers. Only corms with a >3¾ cm diameter are

harvested. Leaves may be kept or discarded depending on whether they are being sold.

After harvest, fields are left in wet fallow before returning to production. The refuge

requires harvested fields to remain in wet fallow for 30 days. A field is sometimes left in

dry fallow after three consecutive crops or if there are signs of disease. Fields are

returned to production by tilling the existing vegetation several times (at deep and then

shallow depths) to incorporate the litter into the soil and flooding the impoundment prior

to planting.

Hawaiian common moorhen ecology

Hawaiian common moorhens are highly sedentary and inhabitat freshwater to

slightly brackish aquatic environments (ESIS 1996a, USFWS 2005). Specific habitat

types used by moorhens include natural ponds, marshes, streams, springs or seeps,

lagoons, and man-made sites such as ephemerally flooded pastures, taro lo’i, lotus

(Nelumbo nutifera) farms, aquaculture (shrimp) ponds, reservoirs, mud-settling basins,

sewage ponds, and drainage ditches (Shallenberger 1977, USFWS 1985, USFWS 2005).

Hawaiian common mooorhen feeding, nesting, and cover sites are usually in close

proximity (USFWS 1985, USFWS 2005). However, specific feeding, loafing, and escape habitats have not been fully described, but the moorhen favors densely vegetated areas

(USFWS 1985, USFWS 2005). Because of its secretive nature, moorhens were the least abundant EWB observed during pilot studies at HNWR (Perkins and Gee, unpublished report).

Figure 1: Hawaiian common moorhen or ‘Alae ‘ula (Gallinula chloropus sandvicensis).

Hawaiian common moorhen nesting has been recorded year-round at HNWR with

a peak between April and June (Perkins and Gee, unpublished report). The vast majority

(95.0%) of nests monitored at the refuge between 1975 and 1979 were in taro lo’i (Byrd

and Zeillemaker 1981). This investigation described plants in nesting habitat but it did not examine habitat characteristics leading to nest-site selection.

Hawaiian coot ecology

Hawaiian coots utilize densely vegetated wetlands adjacent to deep, open

freshwater in lowland areas (ESIS 1996b). Brackish (estuarine) environments are used

although water salinity has not been described. Coots have been observed in taro lo’i,

reservoirs, sewage ponds, cane waste settling basins, old fish ponds, aquaculture

impoundments, broad coastal streams, and occasionally ephemerally flooded pastures.

(Schwartz and Schwartz 1952, Shallenberger 1977, USFWS 1985, USFWS 2005).

Although nonmigratory, Hawaiian coots exhibit pronounced irregular movements

associated with rainfall (Pratt and Bribin 2002). Many Kaua’i birds are assumed to move

from Kaua’i to Ni’ihau in the winter as rains fill intermittent lakes on this inaccessible

island (Pratt and Bribin 2002, USFWS 2005).

Hawaiian coots prefer feeding in open fresh and brackish waters surrounded by

vegetation (Shallenberger 1977). Coots feed near the water’s surface, dive to the bottom

or forage on mud or sand bars (Shallenberger 1977, USFWS 1985, USFWS 2005).

Loafing habitat includes logs, rafts of vegetation, narrow dikes, mud bars, artificial

islands, depressions, and false floating nests (USFWS 1985, USFWS 2005).

Figure 2: Hawaiian coot or ‘Alae ke’oke’o (Fulica alai).

Hawaiian coot nesting has occurred during all months (Shallenberger 1977) with variability during the peak breeding periods (Pratt and Brisbin 2002). Nesting habitat is generally along edges of open water in fairly dense vegetation with some emergent plants, or in small open areas (USFWS 1985, USFWS 2005). Extremely dense clumps of robust emergent plants are avoided by coots (Byrd et al. 1985). The number of coot nests recorded at HNWR has been highly variable and may depend on rainfall (Byrd et al.

1985, Engilis and Pratt 1993).

Hawaiian duck (Koloa) ecology

The Koloa is a bird adapted to a wide variety of aquatic habitats from sea level to

2,500 m (Shallenberger 1977). Specific lowland (below 305 m) freshwater habitat types

include marshes, reservoirs, ephemerally flooded pastures, agricultural lands (including

sugar cane and rice), taro lo’i, mud-settling basins, irrigation ditches, and aquaculture

farms (Swedberg 1967, Shallenberger 1977, Engilis et al. 2002, USFWS 2005).

Occasionally, they inhabit coastal salt evaporation ponds and estuarine areas. Upland

areas include mountain streams within dense evergreen rain forests, stock-watering

ponds, seeps, and subalpine bogs (Swedberg 1967, Shallenberger 1977, USFWS 2005).

Figure 3: Hawaiian duck or Koloa (Anas wyvilliana).

Koloa are more regularly found in lowland wetlands and reservoirs during winter months than summer (Engilis and Pratt 1993), which indicates seasonal movement of birds between breeding and nonbreeding sites. Koloa may also move daily between the lowland wetlands and mountain streams (USFWS 2005).

There is a lack of information regarding the Koloa’s nesting, feeding, and loafing habitat. Koloa may meet these life-history requirements in the same general area or utilize a given area for each of these behaviors depending on habitat conditions

(Swedberg 1967). On Kaua’i, Koloa widely use cultivated taro lo’i and have been observed nesting, loafing, and foraging in the taro lo’i of HNWR (Shallenberger 1977).

They were the most abundant EWB monitored during pilot studies at this refuge (Perkins and Gee, unpublished report).

Koloa broods have been recorded year-round, but are generally concentrated from

December to May (Swedberg 1967). As with other Hawaiian waterbirds, breeding season may be attenuated in response to rainfall patterns; higher periods of rainfall result in prolonged breeding and an increase in population levels (Engilis and Pratt 1993).

Nests are placed on the ground with fetched grasses with bunch-type grasses preferred with overhanging tussocks (Engilis et al. 2002).

Hawaiian stilt ecology

The Hawaiian stilt is associated with fresh, saline, and brackish shallow water environments, primarily in coastal or lowland areas (below 150 m in elevation)

(Robinson et al. 1999). On Kaua’i, stilts are found in foothill impoundments and

wetlands (USFWS 2005). Hawaiian stilts are assumed to move between Kaua’i and

Ni’ihau depending upon wetland habitat conditions (Engilis and Pratt 1993, Reed et al.

1998).

Figure 4: Hawaiian stilt or Ae’o (Himantopus mexicanus knudseni). Foraging areas in fresh water environments include irrigation reservoirs, drainage ditches, sugar-cane settling basins, ponds, taro lo’i, sewage oxidation ponds, ephemeral depressions, flooded pastures, and natural marshes (ESIS 1996c). Brackish foraging habitats include inshore reefs and mudflats, silted beach areas, and breached coastal fishponds. Loafing habitats in all 3 water regimes are open spaces in mudflats, prostrate vegetative mats, pasturelands, and large islands in offshore marine mudflats and fresh or brackish ponds (Shallenberger 1977, USFWS 1985, USFWS 2005).

Hawaiian stilts breed from mid-February through late August with peak nesting

varying among years (Coleman 1981). Nesting habitat includes islets scattered in bodies

of shallow water, clumps of vegetation in flooded areas, barren soil (preferred substrate) protruding from shallow water, floating patches of vegetation, or limestone rock areas

(ESIS 1996c, Robinson et al. 1999). Nesting also occurs on edges of taro lo’i (Broshears

1979) although taro harvesting and deliberate flooding often affect reproduction.

RATIONALE OF STUDY

Biannual statewide surveys indicated that EWB populations have responded positively to wetland management at HNWR (Paveglio et al. 1999, USFWS 2005).

Nevertheless, there is a need for a quantitative and qualitative evaluation of the relative contributions of wetland management and taro cultivation in providing suitable EWB habitat. There is limited information on the response of native and naturalized/exotic wetland vegetation to wetland management on heterogeneous substrates at HNWR.

Furthermore, little is known regarding habitat conditions that occur during the taro agricultural cycle and how each stage in the cycle contributes to the life-history requirements (e.g., foraging, loafing, and nesting) for EWBs. Investigations of vegetation responses to wetland management and taro cultivation practices in a year- round growing season are essential to identify recommendations and refine wetland management and taro cultivation practices (where applicable) to promote recovery of

Hawai’i’s EWBs (Paveglio et al. 1999).

Previous EWB research at HNWR focused on population trends (Shallenberger

1977, Paveglio et al. 1999), nesting ecology (Byrd and Zeillemaker 1981, Asquith and

Melgar 1998), and aquatic invertebrate ecology (Broshears 1979). The most recent pilot studies at the refuge examined the phenology and life-history needs of Hawaiian EWBs

(Perkins and Gee, unpublished report). Pilot studies also developed and tested field- sampling methodologies for waterbird populations and vegetation communities. The goal of this study is to evaluate the roles of wetland management and taro cultivation

practices in providing habitat to meet life-history requirements of EWBs on HNWR.

Specific objectives of this study were the following:

1. To measure the spatial distribution and response (e.g., species richness, visual

obscurity, and vertical structure) of wetland vegetation to management

manipulations (e.g., rototiller treatments) on heterogeneous substrates in

refuge wetlands and taro lo’i to cultivation practices over time.

2. To measure the seasonal abundance and behavior of EWBs within refuge

wetlands and cultivated taro.

3. To characterize vegetation (species richness, visual obscurity, and vertical

structure) within taro lo’i where the highest density of each EWB species

occurred.

4. To use call-response techniques to estimate breeding densities of Hawaiian

common moorhen at HNWR.

5. To characterize vegetation at EWB nest sites (species composition, vegetation

height, visual obscurity, and vertical structure).

6. To measure EWB nesting variables (number of eggs, egg weight, egg length

and width, and number of eggs hatched).

STUDY SITE

HNWR is located on the north shore of Kaua’i County, Hawai’i. The Hanalei

River intersects the 371-ha refuge in its upper and central valley before discharging into the east side of Hanalei Bay (Figure 5). Water discharge from the Hanalei River is higher during the rainy season (November to March) than during the dry season (April to

October). Intensive rainfall events causes periodic flooding of this low-lying flat floodplain and contributes to high productivity by providing adequate water for vegetation, depositing nutrient-rich sediments over the floodplain and neutralizing the acidity of the clay-rich soils (Hue et al. 1997).

HNWR is characterized by forested hillsides, grasslands, ephemerally flooded pastures, taro lo’i, and managed wetlands. The forested slopes of the valley are wooded primarily with exotic species such as guava (Psidium guajava), kukui (Aleurites moluccana), mango (Mangifera indica), silk oak (Grevillea robusta), java plum (Eugenia cuminii), and some native ohia (Metrosideros polymorpha) (Shallenberger 1977). The banks of the Hanalei River are lined with hau (Hibiscus tiliaceus) in the upper reaches and monotypic stands of California grass (Urochloa mutica) and other non-native grasses in the lower valley.

The climate at HNWR is tropical with a normal high January temperature of 26°C and a normal low temperature of 18°C. July’s normal maximum temperature is 29°C with a normal minimum temperature averaging 23°C (National Climatic Data Center).

The average rainfall at HNWR is approximately 200.0 cm/year (Byrd and Zeillemaker

1981).

Hanalei Bay

Hanalei River

N 0 200 400 Meters

Figure 5: Landsat 7 image of Hanalei NWR on January 2003.

Study plots

A total of 290 study plots at HNWR were chosen for this investigation. This

includes 265 cultivated taro lo’i (49.3 ha) and 23 wetland subunits (23.9 ha) managed by

the refuge. During the study, there were only 23 wetland subunits with some active

management (soil disturbance and/or water manipulations) because a functional

distribution and discharge system was lacking or insufficient, personnel was limited, and

21 equipment was inadequate to manage all units intensively. At the start of the study, 13.6 ha of refuge wetlands were overgrown with invasive vegetation. In addition, the water supply system was ineffective because water control structures were of the wrong size, type, and location to facilitate the discharge of excess water from wetlands with 200 cm of annual rainfall. Only 4 refuge wetlands with an area of 3.3 ha had adequate water supply and water control to allow intensive manipulations. Furthermore, personnel to perform manipulations necessary in Hawaiian wetlands were inadequate to treat more than 3.3 ha.

Taro lo’i

Cooperative farmers cultivate 71.2 ha of refuge lands (Table 3) under conditions described within special use permits. These annual permits guide farming activities to provide support for and are compatible with EWB management (Shallenberger 1977).

For example, taro permitees must not harass EWBs and report nesting activities in lo’i.

Table 3: Area farmed for taro by special use permit holders at Hanalei NWR during 2004. Permit holder Area (ha) Diego 7.6 Fitzgerald 10.8 Haraguchi 16.0 Koga 10.8 Legaspi 6.0 Quick 4.4 Spencer 6.0 Watari 6.0 Wong 3.6 Total 71.2

The refuge has 286 taro lo’i (53 ha) ranging from 0.02-0.7 ha in size (not including ditches and dikes).

Taro lo’i are flooded using the refuge’s water distribution system that is composed of a network of ditches, dikes, and control structures. “China Ditch” diverts

water from the Hanalei River at the southern end of the refuge via a concrete pipeline;

water is then carried north above ground along an asphalt road before being diverted

towards the northwestern part of HNWR (Figure 6). This ditch provides irrigation water

to taro lo’i in the southern and northwestern parts of the refuge. “Kuna Ditch” diverts water from “China Ditch” next to the asphalt road near the middle of the refuge. It

travels east under the Hanalei River to the base of bluffs and provides irrigation water to

taro lo’i on the northeastern portion of the refuge. A network of small ditches diverts

water from these primary ditches to taro lo’i throughout the refuge. Water levels within

taro units are manipulated (flooded or drained) by changing riser boards on cement or

metal control structures, by adjusting valves attached to PVC piping or manually

removing soil from the banks of minor ditches.

Maui Lehua is the main taro cultivar grown at the refuge and has a 12-16 month

agricultural cycle from planting to harvest (Perkins and Gee, unpublished report).

Harvesting is accomplished on a sustained-yield basis, thus a taro permitee always has

lo’i in various stages of the agricultural cycle. Habitat conditions prior to harvesting

range from open water to a condition with dense overhead taro cover. Harvesting

removes taro cover leaving habitat conditions with a presence or absence of water and

non-taro emergent vegetation. Taro habitat categories were used to identify

"Kuna Ditch"

"China Ditch"

"Kuna Ditch" "China Ditch"

"Kuna Ditch" diversion from "China Ditch"

"China Ditch"

0 200 400 600 8001000 Meters

Taro lo'i Refuge wetlands N Pasture Primary ditches Secondary ditches Kuhio Highway (#56) Asphalt road Dirt road Hanalei River Hanalei NWR boundary

Figure 6: Distribution of taro lo’i at Hanalei NWR.

habitat conditions associated with the taro agricultural cycle. These habitat categories

include:

1) Recently planted (<25% taro cover) (Figure 7) 2) Early growth or mature (25%-50% taro cover) (Figure 8) 3) Medium to dense growth (>50% taro cover) (Figure 9) 4) Lo’i being harvested (Figure 10) 5) Unvegetated wet fallow (≥ 50% water cover, and <25% emergent vegetation cover) (Figure 11) 6) Vegetated wet fallow (≥50% water cover, and ≥25% emergent vegetation cover) (Figure 12) 7) Unvegetated dry fallow (<50% water cover, and <25% emergent vegetation cover) (Figure 13) 8) Vegetated dry fallow (<50% water cover, and ≥25% emergent vegetation cover) (Figure 14)

Refuge wetlands

HNWR has 3 major wetland units (named “A”, “B”, and “C”), a pond created in

cooperation with Ducks Unlimited (the “DU Pond”), and 4 small impoundments called

the “Rice Mill Ponds”. The 3 major wetland units are divided in an east-west direction

by dikes to create multiple subunits (e.g., wetland unit “C” = subunits “C1”, “C2”, “C3”,

“C4”, and “C5”) (Figure 15). The total area of refuge wetlands on HNWR is 36.0 ha.

Managed wetlands range from 0.3-5.5 ha in size (not including ditches and dikes) for a

total of 29.4 ha (Table 4).

Water management of refuge wetland units requires a system of levees and water control structures to mimic natural hydrological events. “China Ditch” provides water to

“DU Pond” before passing to the south of the “Rice Mill Ponds” and the major refuge

Figure 7: Recently planted (<25% taro cover).

Figure 8: Early growth or mature (25%-50% taro cover).

Figure 9: Medium to dense growth (>50% taro cover).

Figure 10: Lo’i being harvested.

Figure 11: Unvegetated wet fallow (≥50% water cover and <25% emergent vegetation cover).

Figure 12: Vegetated wet fallow (≥50% water cover and ≥25% emergent vegetation cover).

Figure 13: Unvegetated dry fallow (<50% water cover and <25% emergent vegetation cover).

Figure 14: Vegetated dry fallow (<50% water cover and ≥25% emergent vegetation cover).

$T $T "C3"

$T "C2" $T $T $T "C1" $T "C5" "C4" $T $T "B2" $T

"B 6""B 4" "Rice Mills Ponds" "B8" $T $T "B10" "B1" $T"B13" $T "B 3" $T "B7""B5" "B9" $T "China Ditch" "B12""B11" "A1"

"A2-A5" $T Water spillover

0 200 400 600 800 Meters

Refuge wetlands N Taro lo'i Primary ditches Secondary ditches Kuhio Highway (#56) Dirt road Hanalei River Hanalei NWR boundary

Figure 15: Infrastructure of refuge wetland units “A”, “B”, and “C” at Hanalei NWR.

Table 4: Area of refuge wetland impoundments (not including ditches and dikes) at Hanalei NWR during the 2004 field season. Refuge wetland impoundments Area (ha) Wetland subunit “A1” 1.5 Wetland subunits “A2”-“A5” 5.5 Wetland subunit “B1” 1.2 Wetland subunit “B2” 1.0 Wetland subunit “B3” 0.4 Wetland subunit “B4” 0.3 Wetland subunit “B5” 0.3 Wetland subunit “B6” 0.3 Wetland subunit “B7” 0.3 Wetland subunit “B8” 0.3 Wetland subunit “B9” 0.3 Wetland subunit “B10” 0.3 Wetland subunit “B11” 0.3 Wetland subunit “B12” 0.4 Wetland subunit “B13” 0.7 Wetland subunit “C1” 1.6 Wetland subunit “C2” 1.3 Wetland subunit “C3” 1.5 Wetland subunit “C4” 1.4 Wetland subunit “C5” 3.7 “DU Pond” 5.5 “Rice Mill Pond #4” 0.4 “Rice Mill Pond #5” 0.3 “Rice Mill Pond #6” 0.3 “Rice Mill Pond #7” 0.3 Total 29.4

wetlands on the northwestern part of HNWR (Figure 6). The water inlet for “DU Pond”

is located on the south side, but the impoundment drains from the north side. Each of the

“Rice Mill Ponds” has independent inflow on the south side and outflows on the north

side. “China Ditch” loses 5 cfs of its estimated 30 cfs capacity (Wurster, unpublished report) where water spills over the dirt road upstream from the refuge’s major wetland

units (Figure 15). Water is diverted from “China Ditch” by two secondary ditches along

the east side of these wetland subunits. Water inflow is manipulated using control

structures located on the east side of each impoundment. Water outflows are located on

the west side of each impoundment. Subunits “B3” to “B13” are interconnected wetlands

that share a water inflow and outflow from the refuge irrigation system.

Soil characteristics are variable within the refuge’s major units because periodic

flooding has introduced silt and clay to areas that may have once been lined with sandy

beach ridges (L. H. Fredrickson, pers. comm.). In general, refuge units (“A”, “B”, and

“C”) have more coarse-textured soils (sand) to the west and finer-textured soils (silt and

clay) to the east (unpublished data). For instance, the soil in subunit “A1” is

predominantly fine-textured although frequent soil disturbance has resulted in a mix of

coarse and loamy soils (Figure 16). The soils in the southeast half of subunit “B1” are primarily fine-textured while the northwest half are mainly coarse-textured. The soils of subunit “B4” and “B6” are primarily coarse-textured with fine-textured soils in the northeast corner.

Refuge wetland bottoms were contoured to facilitate flooding and drainaging while maintaining variable depths (<50 cm) to promote a diverse, native plant community. For instance, wetland subunit “B12” has the lowest elevation and deepest

water levels; whereas, subunit “B3” has the highest elevation and shallowest water levels.

During my study, the wetland plant community at the refuge was dominated by

invasive California grass, which out-competes native species and eliminates open water,

exposed mudflats or shallow benches (Shallenberger 1977). Barnyard grass

"C3"

"C2"

"C1" "C5" "C4" "B2"

"B 6""B 4" "Rice Mills Ponds" "B8" "B10" "B1" "B13" "B 3" "B7""B5" "B9" "China Ditch" "B12""B11" "A1"

"A2-A5"

0 200 400 600 800 Meters

Refuge wetlands Coarse-textured soil Fine-textured soil N Taro lo'i Primary ditches Secondary ditches Kuhio highway (#56) Dirt road Hanalei River Hanalei NWR boundary

Figure 16: Soils of refuge wetland units “A”, “B”, and “C” at Hanalei NWR (unpublished data).

(Echinochloa crus-galli) is a naturalized plant species that has value as food, cover, and nesting material as well as substrate for invertebrates that are forage for EWBs (USFWS

2005). Prominent native plant communities in refuge wetlands include F. littoralis, nutsedges (Cyperus spp.), and rock bulrush (Schoenoplectus juncoides).

METHODS

Macrohabitat characterization

Refuge wetlands

Spatial and temporal changes in cover of wetland vegetation and water were monitored using a Geographic Information System (GIS). Refuge wetland and taro lo’i boundaries were mapped during pilot studies and georeferenced with Landsat 7 satellite imagery during January 2003. Macrohabitat conditions within refuge wetlands were monitored using a handheld Garmin GPS Map 76 (position accuracy = 5-8 m) to map the area of vegetation and water every 3-6 weeks from late March to late November 2004.

Vegetation sampling of wetland subunits “B4” and “B6” stopped in early fall 2004 while dikes were being reconfigured to reflect differences in soil properties within subunits.

Vegetation communities were mapped only with a canopy cover greater than 25% and measuring at least 3 m in width. The dominant species in each vegetation community was identified and its average height was measured. Data were input into GIS software to assess spatial and temporal changes in cover of wetland vegetation and water.

Taro lo’i

Spatial changes in taro habitat conditions for the 8 categories were recorded weekly within 265 taro lo’i during visual scan surveys for EWBs from early March to early November 2004. The accuracy of taro habitat categories was randomly verified by

measuring taro canopy cover of lo’i within a rectangular sampling frame made of PVC

tubing (50 cm x 100 cm) (Vogel 1999).

Taro cultivation techniques were classed by habitat condition based on the

removal of non-taro emergent vegetation. Each taro permitee on the refuge removes non-

taro emergent vegetation at different rates depending on available manpower; rate of non-

taro growth; and competing demands for planting, fertilizing, and harvesting. Non-taro

emergent vegetation was measured using a 50 cm x 100 cm sampling frame in randomly

selected taro lo’i during the previous pilot study (Perkins and Gee, unpublished report).

Farming intensity for each permitee was classified (intensive or less intensive) based on the percent of sample plots with non-taro emergent vegetation (Table 5) and the

qualitative factors previously mentioned.

Table 5: Percent of sample plots within taro lo’i managed by taro permitees that had non- taro emergent vegetation based on 2003 data. Permitee % of plots with non-taro emergent vegetation Farming intensity (n = number of plots sampled) Watari 2.4% (n = 44) Intensive Koga 4.4% (n = 90) Intensive Haraguchi 8.0% (n = 163) Intensive Diego 20.8% (n = 72 ) Less intensive Wong 25.5% (n = 47) Less intensive Spencer 27.8% (n = 133) Less intensive Legaspi 33.9% (n = 59) Less intensive Quick 36.3% (n = 55) Less intensive Fitzgerald 50.0% (n = 54) Less intensive

Waterbird monitoring

Visual scan surveys

Visual scan surveys were used to determine EWB abundance, behavior, and habitat use of refuge wetlands, taro lo’i, and dikes from 3 March to 12 November 2004.

Scan surveys were conducted from 4 observation points located on the 100 m high bluffs surrounding the middle Hanalei River Valley. Scan surveys were conducted for 4 hours after sunrise and 4 hours before sunset every 3-9 days. Surveys were stopped if precipitation was steady. The order in which sites were visited was randomized for each survey day to minimize bias. A total of 28 morning and 26 afternoon scan surveys were conducted throughout the study.

A Fujinon Super ED 80 spotting scope with 20-60x magnification was used to visually scan for EWBs in wetlands and taro lo’i on the refuge. The general habitat for each observed EWB was characterized by wetland subunit, taro habitat category, and wetland or taro dike. Instantaneous scan surveys noted behavior at first observation and placed birds into broad habitat use and water depth categories within an estimated 4 m radius of the bird (Appendix 1). These data were used to determine habitat use and a time activity budget for each species using the same methods described by Altman

(1974).

The accuracy of waterbird counts is affected by the distance and visibility of individual refuge wetlands and taro lo’i from observation points as well as disturbance of birds from taro cultivation and refuge management activities. Survey sites are up to 1 km from observation points; whereas, others are partially obstructed by trees (Figure 17).

For instance, taro lo’i farmed by Fitzgerald (2.4 ha) near a pasture on the east side of the

refuge were not included in visual scan surveys because these lo’i were too far from

observation point #2 (Figure 18) and completely obstructed by trees from observation

point #3 (Figure 19). Differences in vegetation cover within refuge wetlands and taro lo’i

also limit visibility of EWBs. Waterbirds are frequently seen on dikes. However, taro

farmers mow taro dikes (Figure 20) as frequently as once per week which aids visibility; whereas, infrequent mowing of refuge dikes results in tall grass that decreases the visibility of waterbirds (Figure 21).

Visibility index

The growth of wetland vegetation obscured many EWBs because of the distance

between the point of observation and the unit. To account for this problem, visibility indices were developed by flushing EWBs from refuge wetlands and taro lo’i that were the last to be surveyed during morning or afternoon scans. The visibility index for scan surveys takes into account the vegetation cover (<25%, 25%-50%, >50%) and taro habitat category. However, there was insufficient data on the visibility index to adjust scan survey numbers in the analyses of all EWB species and habitat conditions within refuge wetlands and taro lo’i. Results for the visibility index are reported for Koloa and stilt in taro lo’i because there was sufficient data to give insights into scan survey accuracy.

"Observation point #2" "Observation point #1"

"Observation point #3"

"Observation point #4"

0 200 400 600 800 1000 Meters

Refuge wetlands Taro lo'i N Field of view for observation site #1 Field of view for observation site #2 Field of view for observation site #3 Field of view for observation site #4 Blocked field of view Pasture Kuhio Highway (#56) Asphalt road Dirt road Hanalei River Hanalei NWR boundary

Figure 17: Scan survey observation sites used at Hanalei NWR from 12 January to 12 November 2004.

Figure 18: View of Fitzgerald taro lo’i (in red) from observation site #2 at Hanalei NWR from 12 January to 12 November 2004.

Figure 19: View of Fitzgerald taro lo’i (in red) from observation site #3 at Hanalei NWR from 12 January to 12 November 2004.

Figure 20: Mowed dike surrounding taro lo’i managed by Haraguchi at Hanalei NWR.

Figure 21: Unmowed dike (in red) between refuge wetlands at Hanalei NWR.

Call response surveys

Call response surveys were used to estimate densities of breeding Hawaiian common moorhen because of their secretive nature. A total of 79 surveys sites were monitored, including 60 sites (47.4 ha) located in taro lo’i and 19 sites (15 ha) in refuge wetlands. A disproportional number of sites were surveyed in refuge wetlands as a result of the limited capacity for management during my study.

Surveys were carried out 30 minutes before sunrise and ended 4 hours after sunrise. Survey sites were monitored 3 times separated by 5-12 days (depending on weather conditions) during the peak breeding period from late March until mid-April.

Survey sites were monitored only once during the off-peak breeding period during mid-

October due to time restrictions.

Recordings of the Hawaiian subspecies of the common moorhen (provided by the

Hawaiian Audubon Society) were played on a portable cassette recorder located 0.75 m above the ground (Gibbs and Melvin 1993). Call response surveys lasted 13 minutes because moorhen are more likely to walk out in the open after 10 minutes of observation from dikes (Nagata 1983). The observer recorded spontaneous visual and aural observations of moorhens for 5 minutes after which 1-minute of taped vocalizations were played (Walther and Hohman 1999). Visual and vocalization monitoring continued while playing 10 seconds of the typical call and 10 seconds of the female and chick calling to each other; separated by a 5 second gap. The vocalization sequence was played twice during the 1-minute tape playback (Conway and Gibbs 2005). Moorhen visual and aural observations continued to be recorded for 3 minutes after the tape playback. The 1-

minute of taped vocalizations were then repeated followed by 3 minutes of listening. The

estimated distance, type of broadcast, and habitat type (taro lo’i, refuge wetlands, ditches,

pasture, and riparian habitat) were described for all moorhen observations.

The effective distance at which I could estimate a common moorhen responding

to the call response method was 50 m, giving a circular sample plot of roughly 0.79 ha.

Survey points were placed on dikes at least 100 m apart to avoid double counting and

only moorhens within 50 m of the survey point were noted. The order and time at which sites were surveyed was randomized to minimize bias (Brenowitz 1981, Falls et al. 1990).

Surveys were conducted only when constant wind speeds were < 20 km/h and ceased

with steady precipitation (Ribic et al. 1999). Breeding Bird Survey methodology was

followed for other weather restrictions (Emlen 1971, Emlen 1977).

Microhabitat characterization

Refuge wetlands

The structure and cover (%) of wetland vegetation following wetland

management were quantified to assess the relative contributions of moist-soil

management in satisfying the life-history needs of EWBs on HNWR. A centrical area

sampling design (Lauhban 1992) was used within each stratum of soil texture (fine or

coarse) of a wetland subunit that was rototilled. Each stratum of a wetland was divided

into 0.01-ha grids (10 m x 10 m). A minimum of 10 and up to 35 sample plots were

randomly selected monthly from each refuge wetland depending on size. Because areas

in refuge wetlands with deep water or high elevation were not rototilled, only sample

plots with the rototiller treatment were used in the analysis. Sample plots were randomly

located during each sampling period within each soil strata based on surface soil texture

(fine and coarse).

Visual obscurity, vertical structure, and average number of plant

species/vegetated sample plot were measured every 5-6 weeks following soil disturbance

to determine plant response. Visual obscurity is important as thermal/escape cover for

wildlife (Higgins et al. 1996) and vertical structure supports nests built overwater. Visual

obscurity was measured using a vertical cover pole (3 cm x 150 cm) located at the center

of each sample station grid, and read from a standard distance (4 m) and height (1 m) at

the cardinal directions (Robel et al. 1970). For instance, if the pole is not visible until the

fifth dm, the reading is 4. The cover pole was marked in dm and the height of total visual

obscurity was recorded. Vertical structure of the vegetation next to the pole was

measured in a rectangular sampling frame (50 cm x 100 cm) made of PVC tubing (Vogel

1999). Percent cover was measured for live plants at 3 levels; ground, 15 cm, and

canopy. The percent cover class of live (green) plant species was visually estimated

using modified Daubenmire classes (Table 6). Average number of plant species/vegetated sample plot was calculated for each subunit during each sample period.

Taro lo’i

The structure and density of taro growth were quantified in taro lo’i to assess the

relative contributions of taro in satisfying the life-history needs of EWBs on HNWR. A

total of 8 sample plots representing the 2 taro lo’i with the highest density for each EWB

species were selected following weekly waterbird scan surveys. Each taro lo’i was

divided into 0.01-ha grids (10 m x 10 m). A minimum of 3 and up to 10 sampling frames was measured from each taro lo’i depending on its size.

Table 6: Modified Daubenmire cover classification (Daubenmire 1959). Daubenmire Class range (%) Midpoint (%) cover class 1 0-1 0.5 2 2-5 3.5 3 6-25 14.5 4 26-50 37.5 5 51-75 63.5 6 76-95 84.5 7 96-98 97.0 8 99-100 99.5

The vegetative obscurity and height were measured with a Robel pole (Robel et al. 1970). The vertical structure at the substrate, 15 cm and canopy levels were measured using Vogel’s (1999) protocol. Average number of plant species/vegetated sample plot was calculated for each taro habitat category and level of farming intensity. Microhabitat measurements were not made in unvegetated wet fallow or unvegetated dry fallow taro to avoid disturbing the soil prior to planting. Pooled visual obscurity, vertical structure and average number of plant species/vegetated sample plot were assumed to be zero.

Nest ecology

EWB behavior (e.g., nest building) observed during weekly scan surveys and daily activities were used to locate potential nests in refuge wetlands and taro lo’i.

Systematic searches within refuge wetlands were conducted monthly while georeferencing vegetation communities from late March to late November. Systematic

45 searches within taro lo’i were not possible from 4-6 months after planting until harvesting due to the risk of damaging taro roots by walking among the plants. However, farmers reported moorhen nests they discovered while weeding and harvesting taro, but the frequency and timing of these activities is highly variable for each taro permitee depending upon the intensity of farming. Tape playback of moorhen vocalizations was used to identify breeding moorhen and possible nesting activity in taro lo’i and refuge wetlands from late March to mid-April. These techniques were less effective for Koloa because they utilize dense nesting cover in uplands.

Nest locations were marked (with a bamboo pole or flagging) approximately 4-m away and georeferenced with a handheld Garmin GPS Map76 for monitoring. Eggs were numbered with a soft black felt marker according to the order in which they were laid when possible. Freshly laid eggs were cleanest in appearance. If eggs were fresh, nests were visited every 2-3 days to determine if more eggs were present. Egg length and width were measured and mass was taken every 5-7 days to minimize disturbance to nesting birds. The date of hatching was estimated based upon one or more of the following: date of nest initiation, incubation period, and embryonic stage. Moorhen and stilt eggs also were floated (Alberico 1995) and Koloa eggs were candled (Westerkov

1950) to estimate hatching date.

Nests were classified as successful (≥1 egg hatched) or unsuccessful (no eggs hatched). Eggs were assumed to hatch if they were present in the nest during late incubation, but they were not found in or near the nest after hatching. A search for eggshell fragments and detached membranes was made within a 4-m radius of nests

because adult moorhens have been found to consume shells or carry them away from

nests (Fredrickson 1971). Where possible, efforts were made to determine the fate of

unsuccessful nests (predation, abandonment. or flooding). Nest failure may have been caused by nest monitoring, farmers, or predators. Chicks were monitored during daily activities and weekly waterbird surveys to gain insight into recruitment. All EWB mortalities discovered on the refuge between 22 January and 2 August 2004 were noted and attempts were made to determine the species and cause of death.

Macrohabitat and microhabitat conditions were described for nest sites. The macrohabitat type was classified as taro lo’i, refuge wetlands, ditches or dikes. The farming intensity and taro habitat category were identified for nests in taro lo’i.

Microhabitat conditions surrounding nests were sampled at the time of discovery because of the rapid development of wetland vegetation in Hawai’i. Nesting material was identified and nest measurements included nest length, width, and thickness. The water depth at nests and distance to the nearest dike was measured. The vertical structure was measured using protocols from Vogel (1999). Percent cover was measured for green plants and plant litter at 3 cover strata; substrate, 15 cm, and canopy. All plants in the plot were identified to species and its average height and percent cover (Daubenmire

1959) within the sampling frame was determined. A modified Robel pole technique

(Robel et al. 1970) was used to measure vegetation height and water depth. To assess microhabitat selection for nesting, similar data were collected for an equal number of random sites (dense vegetation, exposed or semi-exposed mudflat, or dike) on the refuge.

Random sites were measured 10 m away because stilts defend 14-30 m around nests

(Robinson et al. 1999). I adjusted the distance so random sites did not fall outside of refuge impoundments.

Data analysis

Macrohabitat characterization

The Spatial Analyst extension for ArcView 3.3 was used to determine the cover of wetland vegetation and water (% of subunit) of all refuge wetlands monitored from late March to late November 2004. Spatial Analyst also was used to estimate the weekly availability (in ha) of each taro habitat category for the average of morning and afternoon

surveys from early March to early November 2004. Because the size of taro lo’i and the

degree of human disturbance were previously identified as limiting factors for use by

Koloa (Engilis et al. 2002), a two-tailed t-test was used to compare the difference in size

between the average taro lo’i for each farming intensity.

Waterbird surveys

Morning and afternoon scan survey numbers were averaged each week to

calculate EWB densities for each refuge wetland and taro habitat category. Two-tailed t-

tests were used to compare the difference between morning and afternoon scan surveys

for daily patterns of abundance. Simple linear regression analysis was used to determine

if seasonal patterns of EWB abundance were related to precipitation (recorded at

Princeville Ranch located less than 1 km northeast of the Hanalei Valley) the day prior to

scan surveys (Appendix 2).

Calculations of EWB densities for microhabitat measurements in taro lo’i assume that waterbirds were using a lo’i if they were observed on the portion of the dike or ditch surrounding it. For instance, a moorhen observed foraging on the edge of a dike next to a lo’i was associated with that lo’i. Furthermore, if that moorhen was observed on a dike, as it was walking between lo’i it was associated with the nearest lo’i in the calculation of moorhen density for measuring taro microhabitat.

Throughout the study, the total area in each taro habitat category available during weekly scan surveys constantly changed. Thus, use of taro habitats was evaluated by comparing the average number of each EWB species per taro habitat category with the total area available for each habitat category during weekly scan surveys. The preference assessment program, PREFER version 5.1, was used to perform calculations described in

Johnson (1980). PREFER tests the hypothesis that all components are equally preferred and compares components using the multiple comparison procedure of Waller and

Duncan (1969) that controls the overall experimental error rate (alpha = 0.05).

Preference assessment with refuge wetlands was not possible during the study because wetland infrastructure was not in place to promote consistent, optimal conditions in all refuge wetlands.

Non-taro emergent vegetation is not easily identifiable in newly planted, early growth or mature, and medium to dense growth taro lo’i using visual scan surveys because of taro canopy cover and the distance from observation points. Thus, comparisons by taro farming intensity (removal of non-taro emergent vegetation) were used in the preference assessement for these taro habitats.

The addition of tape playback of moorhen vocalizations to passive observations was tested for its effectiveness at increasing the number of moorhen detected. The total number of moorhen detected during 5 minutes of passive observation alone was compared to the total number of moorhen detected using 5 minutes of passive observation and 1-minute of taped vocalizations followed by 3 minutes of listening.

Furthermore, the total number of moorhen detected during 5 minutes of passive observation alone was compared to the total number of moorhen detected using 5 minutes of passive observation followed by two consecutive sequences of 1-minute of taped vocalizations followed by 3 minutes of listening.

Multiple call response surveys were tested for its effectiveness at increasing the number of moorhen detected. A detectability index (% detected) for the call response methodology was determined by testing the number of responses at 10 study plots with a known population of 31 moorhens (e.g., study plot has a nesting pair). These sites were surveyed following the final round of call response surveys during the peak breeding season and the detectability index was used to estimate the moorhen population at

HNWR.

Moorhen densities during the peak breeding season were calculated separately for each habitat based on the total area in taro lo’i and refuge wetlands surveyed (e.g., number of survey sites in taro lo’i x area/site = total area surveyed). The detectability index was applied separately to the total number of moorhen responses in taro lo’i and refuge wetlands during the third round of call response surveys (e.g., total number of moorhen responses at survey sites in taro lo’i ÷ detectability index). Binomial expansion

was applied to the detectability index (Zar 1984) to show the improvement in population

estimation by having multiple surveys. Time restrictions prevented a similar test during

the off-peak breeding period.

Microhabitat characterization

Two-tailed t-tests were used to compare microhabitat characteristics (average

number of plant species/vegetated sample plot, pooled visual obscurity, and green cover)

in refuge wetlands between sample periods. A two-tailed t-test also was used to compare

the average size of taro lo’i with the highest densities of each EWB species.

Nesting ecology

Two-tailed t-tests were used to compare microhabitat characteristics (water depth, distance to dike, average number of plant species/vegetated sample plot, pooled visual obscurity, and green cover) of moorhen nests in intensively and less intensively managed lo’i. Moorhen were expected to show preference for nesting in less intensively managed taro lo’i because non-taro emergent vegetation provides additional visual obscurity and structural support. A χ2 goodness-of-fit analysis was performed to determine if there was

a significant difference between the observed and expected number of moorhen nest sites

by farming intensity.

Two-tailed t-tests were used to compare moorhen and stilt nest microhabitat

characteristics for nest success and nest site selection. The Mayfield method (1975) was

used to account for nest and hatchling exposure time in the analysis. The last day a nest

was monitored was taken for exposure time when the exact date of nest failure was

unknown.

Constraints of study

Like most refuges, the geomorphic setting at HNWR has not been fully described.

Specifically, information was lacking regarding soil properties (e.g., texture and

chemistry) as well abundance, distribution, and movement of surface and subsurface water. Because neither a Comprehensive Conservation Plan (CCP) nor HMP have been

completed and because sufficient staff and equipment were not available during my

study, HNWR did not implement refuge wide habitat-based management manipulations during the study. Thus, manipulations to create the diversity of habitat conditions required for all life-history events of the 4 EWBs could not be effectively implemented during my study. Of the 29.4 ha in refuge wetlands, only 3.3 ha were intensively managed for EWBs. These manipulations were specifically designed for stilt nesting and to promote the germination of native wetland vegetation. In contrast, 52.5 of 53.0 ha in taro lo’i were actively farmed and in various stages of the taro agricultural cycle.

Vegetation characteristics within refuge wetlands and taro lo’i were not collected with the same intensity as a result of limited capacity for wetland management during this study. Vegetation within refuge wetlands was monitored monthly on 3.3 ha to evaluate response to wetland management. In contrast, vegetation within taro lo’i was monitored weekly on sites with the two highest densities of each EWB species as determined from scan surveys. Attempts were made to develop visibility indices to adjust for the ability to

detect EWBs within different vegetation structural conditions. Because these efforts were unable to develop visibility indices for all species, analyses of EWB density were based on unadjusted numbers and included birds observed on dikes. Although the most accurate EWB density data were collected from call surveys for moorhen, these surveys were only implemented during 31 March to 23 April 2004.

VEGETATION AND ENDANGERED WATERBIRD RESPONSE TO

ROTOTILLER TREATMENTS

Management strategies using water manipulations and soil disturbance (e.g., rototilling and discing) have been successfully employed to promote conditions that stimulate germination of moist-soil plant species in temperate wetlands (Fredrickson

1996). In contrast, information on the response of tropical wetland native and naturalized/exotic plants to heterogeneous substrates and management actions are lacking on the Hawaiian Islands (Paveglio et al. 1999). One such study was initiated on Kealia

Pond NWR to determine the response of monotypic stands of pickleweed (Batis maritima) to mowing and rototiller treatments (Rader 2005). The objective of my study was to assess the vegetation and EWB response to wetland management at the HNWR.

Results

Vegetation response to rototiller treatments

Tall, dense vegetation and openwater habitats were replaced by low growing

vegetation following rototiller treatments. The perennial, California grass (U. mutica), was most abundant in vegetation communities of wetland subunit “A1” and “B1” prior to treatments (Table 7). This late successional wetland plant often exceeded 100 cm in

height. Early successional wetland plants germinated in mudflats of all wetland subunits

following rototiller treatments and rarely exceeded 50 cm in height. The annual sedge, F. littoralis, was abundant on fine and coarse-textured soils while the perennial knotgrass

(Paspalum distichum), was most abundant on coarse-textured soils that were flooded continuously.

Table 7: Vegetation response after soil disturbance in refuge wetlands at Hanalei NWR during the 2004 field season. Wetland Unit size Dominant vegetation prior Vegetation response subunit (ha) to moist-soil management practice “A1” 1.5 Urochloa mutica Fimbristylis littoralis

“B1” 1.2 Urochloa mutica Paspalum distichum and Fimbristylis littoralis

“B4” 0.3 Mostly open water Paspalum distichum and Fimbristylis littoralis

“B6” 0.3 Mostly open water Fimbristylis littoralis

Plant species richness was high with 16 emergent aquatic plant species sampled in

refuge wetlands following rototiller treatments (Appendix 3). Among the most

frequently sampled species growing on fine and coarse-textured soils were matamat

(Rhynchospora corymbosa), Mexican primrose-willow (Ludwigia octovalvis), E. crus-

galli, and U. mutica. Each of these plants was present in 8.9% to 16.1% of vegetated

sample plots with an average height of less than 30 cm after 46-138 days of growth.

The number of days for the average number of species/ vegetated sample plot to

stabilize following soil disturbance differed depending on soil texture. On fine-textured

soils, the average number of species/ vegetated sample plot was 2.5 (range = 1-6, n =14)

to 3.6 (range 2-8, n = 17) within 89-92 days after initial flooding (Figure 22). In contrast,

the average number of species/ vegetated sample plot on coarse-textured soils was more

variable at 46-138 days (Figure 22) with 1.5 (range = 1-2, n = 4) to 4.0 (range = 2-8, n =

8) species/ vegetated sample plot. There was no significant difference in the average number of species/ vegetated sample plot on subsequent surveys for fine (two-tailed t- test; t = 2.03; d.f. = 35-36; p > 0.05 for all tests) or coarse-textured soils (two-tailed t-test; t = 2.03; d.f. = 8-34; p > 0.05 for all tests).

4 Fine-textured soils

0 0 50 100 150 200 250 5 Coarse-textured soils Wetland subunit "A1" Wetland subunit "B1" 4 Wetland subunit "B4" Wetland subunit "B6" 3 Average number of species number perAverage vegetated plot sample

0 0 50 100 150 200 250 Number of days after initial flooding Figure 22: Average number of species per vegetated sample plot in refuge wetlands “A1”, “B1”, “B4”, and “B6” following soil disturbance and initial flooding at Hanalei NWR during the 2004 field season.

The vegetation cover (% of subunit) in rototilled wetlands was strongly

influenced by the extent of soil surface area without surface water. A partial water

drawdown in wetland subunits “A1” exposed mudflats 12 days after initial flooding. F.

littoralis dominated vegetation communities in wetland subunit “A1” and exceeded 40% cover at 80 days after initial flooding (Figure 23). In contrast, wetland subunit “B6” has

a lower elevation relative to subunit “B4” with both sharing a common inflow and

outflow to the refuge’s water distribution system. The lack of independent water control

in these wetlands resulted in extensive flooding of wetland subunit “B6” where F.

littoralis cover was less than 10% (Figure 23).

The number of days to reach maximum vegetation cover in subunits (% of

subunit) with rototilled treatments after initial flooding was more rapid with F. littoralis

compared to P. distichum. F. littoralis was the dominant species in wetland subunits

“A1”, “B4”, and “B6” where this plant reached maximum cover in 80-86 days (Figure

23). In contrast, P. distichum was dominant in wetland subunit “B1” 87 days after

flooding, but it was eventually overtopped by F. littoralis (which was already present in

the understory) within 135 days after flooding (Figure 23). Vegetation cover reached a

maximum extent in this wetland subunit by 180 days.

Emergent vegetation had relatively low visual obscurity following soil

disturbance, where time to reach maximum visual obscurity varied by wetland subunit.

Maximum visual obscurity was greater than 2.5 dm in wetland subunits “A1”, “B1”,

and“B4” and was reached in 84-138 days after initial flooding (Figure 24). In

100 Fimbristylis littoralis 80

% cover % 40

0 50 100 150 200 250 100 Wetland subunit "A1" Paspalum distichum Wetland subunit "B1" 80 Wetland subunit "B4" Wetland subunit "B6"

% cover % 40

0 50 100 150 200 250 Number of days after initial flooding

“B6” at 46 days after initial flooding.

Visual obscurity fluctuated between sampling periods after reaching its maximum

in most rototilled wetlands (Figure 24). Koloa trampled F. littoralis while feeding and loafing which caused a significant decrease in visual obscurity (two-tailed t-test; t = 1.98; d.f. = 154; p < 0.001) in subunit “A1” at 124 days after initial flooding. P. distichum reached its maximum visual obscurity within 138 days after initial flooding. However,

this creeping grass began to expand horizontally resulting in a significant decrease in

visual obscurity 197 days after initial flooding (two-tailed t-test; t = 2.04; d.f. = 32; p =

0.01). Significant changes in visual obscurity may have been caused by natural water level fluctuations although precise water level data are lacking for these managed

wetlands.

3.0

2.5

2.0

1.5

1.0 Visual obscurity (dm) obscurity Visual Wetland subunit "A1" Wetland subunit "B1" 0.5 Wetland subunit "B4" Wetland subunit "B6"

0.0 0 50 100 150 200 250

Number of days after initial flooding Figure 24: Average visual obscurity (dm) of vegetation sampled in wetland subunits “A1”, “B1”, “B4”, and “B6” following soil disturbance and initial flooding at Hanalei NWR during the 2004 field season. Average percent cover of F. littoralis was greatest at all sampling strata on fine- textured soils of wetland subunit “A1” for all sampling periods except 124 days after initial flooding (Figure 25). This sedge reached a maximum canopy cover of 68.3%

(range = 3.5-97.0, n = 17) and more than 30% cover at lower levels within 89 days after initial flooding. Percent cover of F. littoralis was roughly one-half for fine and coarse- textured soils of wetland subunits “B1” and coarse-textured soils of wetland subunit “B6” compared to subunit “A1”. In contrast, percent cover of F. littoralis on coarse-textured soils on wetland subunit “B4” exceeded 50% after 84 days.

Average percent cover of P. distichum was greater at all levels on coarse-textured soils with continuous flooding compared to fine-textured soils in wetland subunit “A1”

(Figure 26). This perennial grass had a mean canopy cover of 46.1% (range = 14.5-97.0, n = 7) in 89 days on coarse-textured soils and only 5.2% (range = 0.5-14.5, n = 3) after

223 days on fine-textured soils. Average percent cover of P. distichum was similar for both soil textures in wetland subunit “B1”.

Endangered waterbird response to rototiller treatments

The habitat type most commonly used by coot (54.2%), Koloa (53.5%), and stilt

(64.0%) was open water (Appendix 4). These waterbirds were also commonly detected in other emergent vegetation. In contrast, moorhen were most commonly associated with other emergent vegetation (44.8%) and open water areas (31.0%). All EWBs seldom used wet mud and the mud/water interface (Appendix 4).

Moorhen (62.1%), coot (71.3%), Koloa (48.2%), and stilt (82.0%) were most commonly observed foraging following rototiller treatments (Appendix 5). A large portion of moorhen (34.5%), coot (16.9%), and Koloa (17.7%) were detected in

Fine-textured soils Coarse-textured soils 80 80 Canopy Canopy

60 60

40 40 % of sampling frame sampling % of 20 20 frame % sampling of

0 0 0 50 100 150 200 250 0 50 100 150 200 250 60 60 15 cm 15 cm 50 50

40 40

30 30

20 20 % of sampling frame sampling % of % of sampling frame sampling % of 10 10

0 0 0 50 100 150 200 250 0 50 100 150 200 250 40 40 Surface Surface

30 30

20 20 % sampling frame

10 10 % of sampling frame

0 0 0 50 100 150 200 250 0 50 100 150 200 250

Number of days after initial floooding Number of days after initial flooding

Wetland subunit "A1" Wetland subunit "B1" Wetland subunit "B4" Wetland subunit "B6"

Fine-textured soils Coarse-textured soils 70 70 Canopy Canopy 60 60

50 50

40 40

30 30 % of sampling frame sampling % of 20 20 frame sampling % of

10 10

0 0 0 50 100 150 200 250 0 50 100 150 200 250 20 20 15 cm 15 cm

15 15

10 10 % of sampling frame sampling % of % sampling frame of 5 5

0 0 0 50 100 150 200 250 0 50 100 150 200 250 25 25 Surface Surface 20 20

15 15

10 10 % of sampling frame sampling % of % of samlping frame samlping % of

5 5

0 0 0 50 100 150 200 250 0 50 100 150 200 250

Number of days after initial flooding Number of days after initial flooding

Wetland subunit "A1" Wetland subunit "B1"

locomotion which is also an indication of feeding activity. Foraging activities were

especially common in rototilled wetlands 17-23 days after initial flooding. Furthermore,

20.4% of Koloa were observed loafing predominantly in wetland subunit “A1” after F.

littoralis had reached its maximum cover and height.

Coot (52.7%) and Koloa (43.8%) were commonly detected swimming (59.9%).

Stilt (71.4%) were consistently detected in water up to the joint. Moorhen were detected

in a variety of water depths including water up to the joint (24.1%), wet mud (17.2%),

and swimming (13.8% - Appendix 6). Water depth could not be determined for 20.7% of

coot and 24.1% of moorhen observations due to dense emergent vegetation.

Furthermore, 19.7% of Koloa observations were in dry mud because water was drawn down in wetland subunits “B4” and “B6” in preparation for a reconfiguration.

Discussion

Vegetation response to wetland management

Refuge staff can use soil disturbance and water level manipulations to setback

wetland succession by creating mudflats for the germination of wetland vegetation.

These mudflat conditions resulting from tillage treatments produced bare soil where the dark surface tends to produce warm conditions where the soils dry during a drawdown.

This drying reduces the water in the pore spaces and allows germination of seeds that cannot occur in anoxic conditions. Shallow reflooding following the rototilled treatment eliminated undesirable upland plants that are not tolerant of saturated conditions. The addition of water also tends to cool the soil.

F. littoralis requires fluctuating temperatures for germination (Pons and Schröder

1986) which can be achieved by intermittent flooding to a depth of 8 cm to stimulate seed

germination (Kent and Johnson 2001). F. littoralis response (vertical structure) to

management was strongest on fine-textured soils in wetland subunit “A1”. In contrast, F.

littoralis response (vertical structure) to treatments was slower on coarse-textured soils of wetland subunit “B1” where there was competition from P. distichum.

P. distichum reproduces from rhizomes, stolons, and seeds (USDA 2000).

Mowing vegetation prior to the rototiller treatment in wetland subunit “B1” may have

created conditions for stolon development of P. distichum, a condition that encourages a

denser sod. However, P. distichum is a shade intolerant grass (USDA 2000) that is

unlikely to grow beneath the tall, dense vegetation present in this wetland prior to setting

back succession through management. P. distichum seeds yield their highest germination

when temperatures are between 28 to 35º C and when they receive about 16 hours of light

(USDA 2000). The average maximum daily air temperature during soil disturbance fell into the lower end of this range while 12-13 hours of daylight was available for seed germination (National Climatic Data Center). Mudflats may have produced suitable

conditions to stimulate germination of P. distichum seeds.

Endangered waterbird response to wetland management

Broshears (1979) suggested that tilling a flooded lo’i dramatically increases the

availability of chironomid larvae and other benthic invertebrates which attract foraging

Hawaiian stilts. Soil disturbance from tilling promotes massive amounts of chironomid

larvae with a 2-week growth period. Similarly, EWBs were most commonly observed

foraging in open water areas 17-23 days after soil disturbance and initial flooding of

refuge wetlands. An on-going invertebrate study at the refuge will assess the relationship

between foraging activities by EWBs and aquatic invertebrate abundance following soil

disturbance in wetlands. My nesting study at the refuge has determined the suitability of

soil disturbance at producing nesting habitat for breeding stilts. Soil disturbance in

refuge wetlands enhanced growth of moist-soil plants. These wetland plants provided

structure for aquatic invertebrates and, in turn, increased availability of forage for coot,

Koloa, and moorhen.

Moist-soil plants growing following rototiller treatments are suited for foraging coot, Koloa, and moorhen while providing structure for aquatic invertebrates. Seed production of F. littoralis was abundant in wetland subunit “A1” 94 days after the rototiller treatment (pers. obs.). Coot, Koloa, and moorhen were commonly observed foraging in this wetland within 10-16 days of flooding above the height of F. littoralis seed heads. In contrast, coots were the only waterbird observed feeding on P. distichum although this creeping grass provides structural habitat for invertebrates when flooded

(Murkin 1989).

Each seed of F. littoralis weighs roughly 2.6 mg (Pons and Schröder 1986) and each plant can produce up to 10,000 seeds (Chin et al. 2001). Seed production has been estimated using seed number, seed weight, and stem counts for the early successional plant, E. crus-galli (Laubhan 1992, Laubhan and Fredrickson 1992). Attempts should be made to develop regression equations to estimate seed production of other early

65 successional plants such as F. littoralis after rototiller treatments and over subsequent years. Estimating seed production for early and late successional plants also has benefits for EWB management in taro lo’i because seeds of non-taro emergent vegetation provide nutrients needed by EWBs for successful molting and reproduction.

Estimates of E. crus-galli seed production on the mainland range from over 2,800 kg/ha in early stages of succession to less than 600 kg/ha in later stages of succession

(Fredrickson and Taylor 1982). E. crus-galli in later stages of succession produced 506-

700 kg/ha on fine-textured soils at HNWR; whereas, seed production in refuge wetlands with coarse-textured soils ranged from 672-965 kg/ha (Perkins and Gee, unpublished data). With the reconfiguration of the 24 ha of wetlands on HNWR in 2006, intensive management is now possible on 12 ha of coarse textured soils. If about half of this area

(6 ha) were managed intensively for E. crus-galli annually, there would be the potential to produce 1,400 kg/ha as a conservative estimate. Thus a total of 8,400 kg of E. crus- galli could be produced. The value of E. crus-galli seed production for Koloa can be estimated by using the nutritional requirement using the nutritional requirement of 0.1 kg/duck/use day for Mallards (Anas platyrhynchos) as a surrogate (Reinecke et al. 1989).

Thus, 8,400 kg of E. crus-galli seeds would provide about 84,000 use days for Koloa or

230 Koloa could meet their daily requirements during a year. This is a conservative estimate of the value of intensive moist-soil management on HNWR because Koloa are roughly half the body mass of Mallard (Drilling et al. 2002, Engilis et al. 2002). E. crus- galli also has other benefits other than producing seeds. The robust structure provides

66 visual barriers and nesting structure for EWBs, and most importantly the residual vegetation provides invertebrate habitat.

ENDANGERED WATERBIRD RESPONSE TO SEASONAL CHANGES IN

TARO HABITAT AVAILABILITY

Information on the spatial and temporal distribution of taro habitat types

throughout a year-round growing season is lacking for the refuge. Studies are needed to

establish the links between habitat conditions that occur during the taro agricultural cycle and how each stage in the cycle relates to meeting the life-history requirements of EWBs at HNWR.

Results

Taro habitat availability

The area, distribution, and size of refuge taro lo’i differ for both levels of taro

farming intensity. A total of 49.3 ha of the refuge’s taro lo’i were observable during

weekly scan surveys with intensive and less intensive managed lo’i having an area of

27.4 ha and 21.9 ha respectively. All intensively managed taro lo’i were located in the northern portion of the refuge; whereas, most less intensively managed lo’i were in the southern portion of the refuge. Furthermore, there was a significant difference (two- tailed t-test; t = -8.76; d.f. = 262; p < 0.01) in size between intensively ( = 0.25 ha, range

= 0.05-0.8 ha, n = 156) and less intensively managed taro lo’i ( = 0.14 ha, range = 0.02-

0.4 ha, n = 108).

The length of time a lo’i is in each taro habitat category during the taro

agricultural cycle is as follows: recently planted taro 4-6 weeks; early growth taro 6-8

weeks and mature growth taro 5-9 weeks; medium to dense growth taro 20-36 weeks;

taro being harvested 1-4 weeks; unvegetated or vegetated wet fallow taro up to 4 weeks;

unvegetated dry fallow taro up to 1 week; and vegetated dry fallow taro up to 12 weeks.

The area in each taro habitat category constantly changed from the time of planting through harvest and the fallow period (Appendix 7). During March to October, the area in medium to dense growth taro decreased as early or mature growth taro increased (Figure 27). The area of taro harvested and planted increased from June to

September. Area of unvegetated wet fallow taro decreased in area between June and

August. In contrast, the area of unvegetated wet fallow taro decreased and area in

unvegetated dry fallow taro increased between June and August.

30 Recently planted Early or mature growth Medium to dense growth 25 Lo'i being harvested Unvegetated wet fallow Vegetated wet fallow Unvegetated dry fallow 20 Vegetated dry fallow

15 Area (ha) Area

0 5 9 7 7 5 0 9 4 2 8 4 1 6 1 7 6 2 8 5 3 0 5 2 6 3 0 8 5 2 / / 1 2 1 / 1 / / l 1 / 2 1 2 2 e / 2 1 2 3 . 1 1 2 3 . 1 1 3 8 i 1 y y y . . . . t . . . . 5 6 r il 7 il y y y n 6 e l l y y y g t t t c t r r p r r y u u l l l u g v 1 2 1 u 1 n J J p p p O c . . p l p Ma J u u u u A u e e e o r r A i Ma Ma Ma e J J J J O Ma Ma A r A Ma n A S S S N p u Ma Ma A J

Figure 27: Average area (ha) in taro habitat categories during morning and afternoon surveys at Hanalei NWR from 3 March to 12 November 2004.

Waterbird use of taro lo’i

Daily and seasonal patterns of abundance

The number of observations of each EWB species in refuge taro lo’i was highly

variable. Koloa were the most frequently detected EWB in refuge taro lo’i with 7776

observations while coots were detected less than other EWBs with only 775 observations

(Appendix 8).

Coot and Koloa showed variation in daily patterns of abundance with greater

numbers in afternoon versus morning surveys. The number of coot detected in mornings

( = 8.0 birds, range = 0-32, n = 35) and afternoons ( = 14.8 birds, range = 0-54, n =

33) was significantly different (two-tailed t-test; t = 2.00; d.f. = 66; p = 0.01).

Furthermore, there was a highly significant difference (two-tailed t-test; t = 2.00; d.f. =

66; p < 0.001) in the number of Koloa detected in morning ( = 87.7 birds, range = 30-

187, n = 35) and afternoon surveys ( = 142.5 birds, range = 50-227, n = 33).

EWBs showed three patterns of seasonal abundance in taro lo’i between March

and November. Numbers of moorhen and coot had an upward trend from March to

November (Figure 28). Koloa numbers were highest from January until early May;

whereas, stilt numbers were highest from early May until November (Figure 28).

Seasonal patterns of Koloa numbers in taro lo’i were moderately correlated (R2 =

0.29; F 1, 66 = 2.00; p = 0.16) with rainfall on the day preceding surveys. However,

rainfall on the day preceding surveys was not significantly correlated with seasonal

2 2 patterns of moorhen (R = 0.0035; F 1, 66 = 0.28; p = 0.628), coot (R = 0.0024; F 1, 66 =

2 1.63; p = 0.206), and stilt abundance (R = 0.0002; F 1, 66 = 0.014; p = 0.91) in taro lo’i.

200 Hawaiian common moorhen 180 Hawaiian coot Hawaiian duck 160 Hawaiian stilt

140

120

100

Average ofandmorning afternoon surveys 20

3 3 7 /4 0 7 4 /5 /9 7 7 5 0 9 4 /2 8 4 1 6 1 7 6 2 8 5 3 0 5 2 6 3 0 8 5 2 1 2 2 1 1 2 1 2 l 1 1 2 1 2 2 e 1 2 1 2 3 . 1 1 2 3 1 1 / / / 3 / / / 3 8 / / i / 1 / g ...... 2 0 6 . 9 6 3 . . 5 6 r il 7 il ay n 6 e ly ly y y y t t t ct. 1 2 2 . 1 2 1 2 p r 1 r y ay ay ay u 1 n u u l l l u g ct v ar ar M J J J u u u u O o . . . eb b . . . . A p l p u J J J A ep ep ep e r i M M M e J A O N F b b M M a ar A r A Ma n S S S an an an F e e p u J J J F F M M A J

Figure 28: Endangered waterbird average numbers during morning and afternoon surveys at Hanalei NWR from 12 January to 12 November 2004.

Endangered waterbird use of taro habitat

Birds present on dikes surrounding the 49.3 ha of taro lo’i under observation were highly visible because taro farmers maintain dikes by mowing. Observations on taro dikes accounted for 70.4% of Koloa, 67.9% of moorhen, and 37.9% of coot detected

(Appendix 8). Stilt were detected on taro dikes only 10.9% of the time.

EWBs present in taro lo’i (excluding dike observations) were most frequently detected (percentage of observations) in open taro habitats with high visibility. Stilts

(35.9%), coot (20.3%), and Koloa (10.7%) were most frequently observed in unvegetated

wet fallow (Appendix 8). Moorhen were most frequently observed in lo’i being

harvested (8.2%) and in unvegetated wet fallow (7.9%). Stilt (25.8%) also commonly

used recently planted lo’i; whereas, coot (16.9%) and Koloa (4.9%) also were frequently

observed in lo’i being harvested.

Behavior of endangered waterbirds in taro habitat

Birds present on dikes were most frequently observed (percentage of

observations) foraging, preening or loafing. The percentage of time spent foraging in

each taro habitat varied by species. Foraging accounted for 30.2% of moorhen

detections; whereas, 37.4% of coots were observed preening on taro dikes (Appendix 9-

10). Loafing on dikes accounted for 73.6% and 37.1% of Koloa and stilt detections

respectively (Appendix 11-12). EWBs present within taro lo’i were most commonly

observed foraging in all taro habitats with the exception of Koloa and stilt loafing in

unvegetated dry fallow (Appendix 11-12).

Distribution of waterbirds in relation to water depth in taro lo’i

Shallowly flooded taro lo’i had water depths for EWBs suitable for wading or

swimming. Moorhen were most frequently observed (percentage of observations) in

water up to the joint or deep enough for swimming in flooded taro lo’i (Appendix 13).

Coot and Koloa in flooded taro lo’i were most frequently observed in water deep enough

for swimming (Appendix 14, 15). Stilt were most frequently detected in water up to the joint and occasionally deep enough for swimming in flooded taro lo’i (Appendix 16).

EWBs may have been attracted to moist conditions in dry fallow taro lo’i to feed

on invertebrates available on mudflats exposed by water drawdowns. Moorhens, Koloa, and stilt were observed most frequently in wet or dry mud in unvegetated dry fallow

(Appendix 13, 15, and 16); whereas, most coot observations were on sites with shallow water at the foot (Appendix 14). In vegetated dry fallow, 43.6% of stilts were detected at low lying sites in water depths at the foot; whereas, Koloa utilized water deep enough to swim 24.0% of the time.

Endangered waterbird densities in eight taro habitat categories

EWB densities for each taro habitat (excluding dike observations) fluctuated with canopy cover during the taro agricultural cycle. EWB densities were highest during harvesting and in unvegetated and vegetated wet fallow lo’i (Table 8). The removal of

water from lo’i during wet fallow stages resulted in lower EWB densities in dry fallow

lo’i (Table 8). Stilt were the only species with relatively high densities in the open water

habitat of recently planted taro lo’i (Table 8). As canopy cover increased, densities of

EWBs decreased with the exception of moorhens which tended to increase their use

(Table 8).

Endangered waterbird preference of eight taro habitat categories

Each EWB species preferred certain of the eight taro habitat categories (F 7, 21 =

8.85-62.80; p < 0.001 for all tests). Hawaiian coot preferred vegetated wet fallow lo’i

first; whereas stilts preferred unvegetated wet fallow first (Table 9). Moorhen preferred

taro lo’i being harvested first and vegetated wet fallow lo’i second while Koloa preferred

vegetated wet fallow lo’i first and unvegetated wet fallow second (Table 9).

Table 8: Average endangered waterbird density (birds/ha ± SD) for 28 weekly surveys by taro habitat category at Hanalei NWR from 3 March to 12 November 2004.

Recently planted taro Early growth mature or Medium to dense growth Lo’i being harvested Unvegetated wet fallow Vegetated wet fallow Unvegetated dry fallow Vegetated dry fallow Hawaiian 0.11 ± 0.32 ± 0.01 ± 0.37 ± 0.32 ± 0.36 ± 0.08 ± 0.08 ± common 0.18 0.09 0.05 0.05 0.49 0.90 0.20 0.25 moorhen

Hawaiian 0.03 ± 0.03 ± 0.01 ± 0.21 ± 0.18 ± 0.12 ± 0.01 ± 0.04 ± coot 0.10 0.09 0.04 0.63 0.60 0.47 0.11 0.40

Hawaiian 0.25 ± 0.03 ± 0.01 ± 0.69 ± 1.19 ± 1.14 ± 0.08 ± 0.20 ± duck 0.57 0.11 0.04 1.06 1.71 1.22 0.28 0.59

Hawaiian 0.83 ± 0.07 ± 0.02 ± 0.56 ± 2.20 ± 0.66 ± 0.23 ± 0.06 ± stilt 0.91 0.13 0.05 0.68 3.44 1.44 0.66 0.25

Endangered waterbird habitat selection by taro farming intensity

Preference for eight taro habitat categories based upon the two levels of farming intensity were different for the four EWBs (F 1, 27 = 4.50-126.00; p ≤ 0.05 for all tests).

Moorhen preferred less intensively managed compared with intensively managed lo’i in early growth or mature first and medium to dense growth second (Table 10). Coot preferred less intensively managed compared with intensively managed taro lo’i in medium to dense growth first (Table 10). Koloa preferred less intensively managed compared with intensively managed lo’i from planting until the start of harvesting (Table

10). Stilt tended to prefer intensively managed compared to less intensively managed lo’i being harvested (Table 10).

74 2 8 7 4 1 3 6 5

) Rank of preference (Johnson 1980 2.16 2.16 2.05 0.98 0.77 -1.38 -1.38 -0.66 -2.73 -1.19 Hawaiian stilt Hawaiian common common Hawaiian Average Average difference in ranks 7 5 4 3 2 1 8 6

) Rank of preference (Johnson 1980 0.53 0.53 0.21 0.04 1.89 0.48 -0.64 -0.64 -1.16 -1.36 Hawaiian duck Average Average difference in ranks 8 5 7 2 3 1 6 4

) Rank of preference (Johnson 1980 1.50 1.50 0.29 1.14 1.14 -1.05 -1.05 -0.86 -1.88 -0.43 Hawaiian coot Average Average difference in ranks

6 4 5 1 3 2 8 7

) i NWR from 3 March312 from November to 2004.i NWR Rank of preference (Johnson 1980 0.44 0.44 0.16 1.57 1.57 -0.02 -0.02 -1.23 -0.59 -0.89 Hawaiian comonmoorhen Average Average difference in ranks

Recently planted Recently planted Early growth or mature dense to Medium growth being Lo’i harvested wet Unvegetated fallow Vegetated wet fallow dry Unvegetated fallow Vegetated dry fallow Habitat categories in categories Habitat lo’i taro Table 9:Table Average differenceranks preference and in ofhabitat taro categoriesusage by and availability for thatfor category moorhens, coots, ducks, and stilts at Hanale

75 1 2 2 1 2 1 1* 2*

) Rank of preference (Johnson 1980 ween farming farming ween 0.23 0.23 0.13 0.13 Hawaiian stilt -0.23 -0.23 -0.13 -0.13 0.054 0.054 0.089 0.089 -0.054 -0.089 farming intensityfarming for Average Average difference in ranks 1 2 2* 1* 2* 1* 2* 1*

) Rank of preference (Johnson 1980 0.41 0.41 0.27 0.21 -0.41 -0.41 -0.27 -0.21 0.036 0.036 -0.036 Hawaiian duck Average Average difference in ranks 2 1 2 1 1 2 2* 1*

) Rank of preference (Johnson 1980 R from 3 March12R from 3 November to 2004(*significant bet difference 0.13 0.13 0.21 0.20 -0.13 -0.13 -0.21 -0.20 0.036 0.036 Hawaiian coot -0.036 bitat categories by usage and availability for that category by Average Average difference in ranks 2 1 1 2 2* 1* 2* 1*

)

Rank of preference (Johnson 1980 moorhen 0.21 0.21 0.36 0.27 -0.21 -0.21 -0.36 -0.27 0.036 0.036 -0.036 Hawaiian common Average Average difference in ranks

Intensive Less intensive Intensive Less intensive Intensive Less intensive Intensive Less intensive Farming intensity 0.05). 0.05). ≤ Recently planted Early growth or mature Medium to dense growth Lo’i being harvested Table 10: Average difference in ranks and preference of taro ha Hawaiian common moorhens, coots, ducks,and stilts at Hanalei NW intensities; p Habitat categories in taro lo’i

Vegetation characteristics of taro habitat

Plant species richness was relatively high with a total of 31 species of aquatic vegetation sampled in taro lo’i with the highest observed densities of each EWB

(Appendix 17). Taro was the dominant canopy cover in lo’i prior to harvesting. The main non-taro emergent species were marsh seedbox (Ludwigia palustris), blunt spike

rush (Eleocharis obtusa), S. juncoides, A. coccinea, heartshape false pickerelweed

(Monochoria vaginalis), F. littoralis, and variable flatsedge (Cyperus difformis). Each of these low growing plants (average height = less than 30 cm) was found in 8.9 to 16.1% of vegetated sample plots. Minute duckweed (Lemna perpusilla), a floating plant, was found in 27.4% to 36.2% of vegetated sample plots.

The average number of species/vegetated sample plot differed during the taro agricultural cycle and between EWB species. The number of species/vegetated sample plot increased in taro lo’i after harvesting because non-taro emergent vegetation is no longer removed (Appendix 18). The highest number of species/vegetated sample plot was found in vegetated wet or dry fallow lo’i with such species as C. difformis, manyspike flatsedge (Cyperus polystachyos), Echinochloa crus-galli, Eleocharis obtusa, forked fimbry (Fimbristylis dichotoma), F. littoralis, shortleaf spikesedge (Kyllinga brevifolia), and S. juncoides being common. Furthermore, taro lo’i with the highest densities of moorhen or Koloa tended to have more plant species than lo’i with the highest densities of coots or stilts.

Moorhen and coot had greater requirements than Koloa and stilt for visual obscurity in certain taro habitat categories. Visual obscurity was highest for moorhen in

vegetated dry fallow lo’i (Appendix 19); whereas, visual obscurity was higher for coot

compared to other EWBs in recently planted and medium to dense taro.

There were no apparent patterns to interspecific differences in percent cover at all

strata (vertical structure) for certain taro habitat categories. Plant cover at all strata was

greatest for Koloa in early growth or mature and vegetated wet fallow lo’i (Appendix 20).

Plant cover at all strata was highest for stilt (Appendix 21) and lowest for moorhen in lo’i

being harvested (Appendix 22).

Table 11: Hawaiian duck and stilt visibility index (number in survey/number in walk-by) by taro habitat category at Hanalei NWR from 3 March to 12 November 2004. Hawaiian duck Hawaiian stilt Habitat categories Number of Visibility Number of Visibility in taro lo’i walk-bys index walk-bys index Recently planted 7 81.8% (54/66) 12 67.9% (19/28) Early growth or mature 13 0.0% (0/36) 5 0.0% (0/8) Medium to dense growth 14 0.0% (0/40) 6 0.0% (0/17) Lo’i being harvested 6 32.0% (8/25) 4 75.0% (9/12) Unvegetated wet fallow 4 0.0% (0/7) 8 61.9% (13/21) Vegetated wet fallow 7 57.1% (28/49) 3 66.7% (4/6) Unvegetated dry fallow 1 0.0% (0/1) 1 100.0% (2/2) Vegetated dry fallow 1 0/0% (0/8) 1 0.0% (0/22)

Visibility index

There was sufficient data for calculating a visibility index for Koloa and stilts in

all taro habitat categories with the exception of unvegetated and vegetated dry fallow taro

lo’i. The visibility index was 0.0% for early or mature growth and medium to dense

growth taro lo’i for both species. The visibility index for stilts during harvest (61.9%)

and wet fallow (66.7%) stages was similar to recently planted taro lo’i (67.9%; Table 11).

The visibility index for Koloa was lower than stilts in taro lo’i after harvesting began

because Koloa likely moved to taro lo’i sampled after completion of scan surveys.

Discussion

Taro habitat availability

The length of time a lo’i is classed in each stage of the taro agricultural cycle is

dependent on farming practices and environmental factors. Farming practices that

influence taro cover include planting density (De la Pena 1977, Liou 1984), type of

cultivar used (De la Pena 1983), and degree of nitrogen fertilization (De la Pena 1972,

Manrique 1994). Environmental factors affecting taro growth include temperature

following planting (Lu et al. 2001), cloud cover (De la Pena 1983), soil texture and

chemistry (Manrique 1994), and water availability (De la Pena 1983). Therefore, the length of time a lo’i is classed in each stage of the taro agricultural cycle is difficult to

predict.

Waterbird use of taro lo’i

Daily and seasonal patterns of abundance

Koloa (Engilis et al. 2002) and American coots (McCracken 1987) are wary of

human disturbance. On the refuge, human disturbance from taro cultivation was higher

79 during mornings and may have resulted in lower abundance of Koloa and Hawaiian coots during visual surveys. Koloa in early growth or mature and medium to dense growth taro lo’i not observed during morning scan surveys were frequently observed during walk-bys

(pers. obs.). More data are needed to produce a comprehensive coot visibility index in taro lo’i.

Engilis and Pratt (1993) suggested that Koloa moved seasonally between breeding sites during the summer and nonbreeding sites in lowland wetlands and reservoirs on

Kaua’i during winter months. Koloa seasonal abundance in taro lo’i showed a pattern that supports this theory. Koloa abundance was positively correlated with rainfall suggesting that seasonal movements may be a response to differences in rainfall between the wet and dry seasons. A satellite radio-telemetry study might enhance understanding of seasonal movements of these waterbirds.

Hawaiian stilts (Engilis and Pratt 1993, Reed et al. 1998) and coots (Engilis and

Pratt 1993, Pratt and Bribin 2002) are assumed to move from Kaua’i to Ni’ihau as ephemeral ponds fill with seasonal rains. A sudden and sharp increase in stilt abundance in taro lo’i at the onset of the dry season (early May) supports this theory. Patterns of coot abundance in refuge taro lo’i showed a steady rise from January through November although seasonal patterns of coot abundance are best examined in refuge wetlands where they occur in larger numbers. Stilt and coot numbers rose sharply in refuge wetlands during early May indicating birds are arriving from sites outside of HNWR. However, this theory cannot be substantiated with empirical evidence because neither waterbird surveys from other parts of Kaua’i nor weather data from Ni’ihau are available.

Hawaiian (Nagata 1980, Engilis and Pratt 1993) and Mariana common moorhen

(G. chloropus guami; Takano and Haig 2004) move in response to droughts and wet

periods and readily disperse to seasonal wetlands within islands. Flooded taro lo’i are

available year-round at the refuge which would limit dispersal by Hawaiian common

moorhen. Call response surveys may give indications that moorhen move between taro lo’i and refuge wetlands as suitable foraging habitat becomes available.

Endangered waterbird use of taro habitat

Geese preferred habitats grazed by cattle (Chabreck 1968, Stutzenbaker and

Weller 1989) because the regrowth had higher nutritive value, and also because the short vegetation provides better visibility and feeding efficiency (Kadlec and Smith 1989).

Frequent mowing of dikes surrounding taro lo’i produces many of the same benefits for foraging coots, moorhens, and Hawaiian geese. Given the relatively uniform topography within flooded taro lo’i, dikes are the only dry sites available for EWB loafing and preening. The large portion (73.6%) of Koloa observed loafing on dikes adjacent to taro lo’i during scan surveys suggested the value of this habitat for these behaviors.

Optimum nesting habitat for breeding American coots and common moorhens (G. chloropus) is generally an equal proportion of dense emergent vegetation and open water

(Weller and Fredrickson 1973). Because taro has greater than 25% cover (early growth or mature and medium to dense taro lo’i) for most of its agricultural cycle (Perkins and

Gee, unpublished report), 29.3-29.7 ha of taro habitat had suitable nesting conditions for coots and moorhens year round. Because the visibility index for Koloa and stilts is extremely low in early growth or mature and medium to dense taro lo’i due to high visual

obscurity, EWBs are more easily observed in newly planted, lo’i being harvested, and fallow lo’i.

Distribution of waterbirds in relation to water depth in taro lo’i

Water depths in shallowly flooded taro lo’i on the refuge are relatively uniform and rarely exceed 15 cm (Perkins and Gee, unpublished data), which is suitable habitat for all foraging EWBs. Hawaiian stilts require water depths of 13 cm or less for optimal foraging (Telfer 1973); whereas, Koloa forage in water depths of 24 cm or less (Engilis et al. 2002). Coots forage in water less than 30 cm; whereas, moorhen habitat is characterized by water less than 100 cm deep (USFWS 2005).

Endangered waterbird taro habitat selection

Stilt use of taro lo’i increased sharply following harvest and remained steady

throughout the wet fallow period. Stilt use declined when lo’i were replanted and during

the early stages of taro growth. After a taro canopy cover developed, stilts were no

longer present in lo’i. Broshears (1979) described a strong positive correlation between

stilt and chironomid larvae abundance with similar patterns of distribution relative to the

taro agricultural cycle. Chironomid larvae were scarce in lo’i with full taro canopy cover

but large chironomid populations were established following harvest. My study showed

a similar pattern for Koloa and stilt densities indicating these birds utilized invertebrate

prey items that became established during the wet fallow period. In contrast, coot and

moorhen densities were highest during harvesting and remained high during the wet

fallow period.

Coots and moorhens are mostly herbivorous rails during most of the year (Bannor

and Kiviat 2002, Brisbin and Mowbray 2002) and may forage primarily on seeds of

annual plants, duckweed, and algae under the taro canopy. During the breeding season,

aquatic invertebrates are more important and account for a greater proportion of foods in

the diet (Desrochers and Ackney 1986, Driver 1988). Furthermore, they forage

opportunistically on aquatic invertebrates and egg masses released in the water column

during harvesting. Removal of taro plants as a result of harvesting increased the visibility of these secretive birds during scan surveys. These rails along with Koloa can feed on the seeds of annual plants in vegetated wet fallow lo'i. In fact, coots and Koloa preferred vegetated wet fallow taro lo’i over other taro habitat categories.

Broshears (1979) proposed that the chironomid population response in wet fallow lo’i is a result of the combination of massive egg layings following taro harvest with a larval growth period of 2 weeks. Because adult chironomids attach their eggs to vegetation and other bottom-anchored debris (Williams 1944, Hardy 1960, Murkin 1989) they tend to be abundant in freshly harvested and vegetated wet fallow taro lo’i. Taro harvesting increases during summer to meet peak commercial demand (NASS 2005), thus availability of wet fallow lo’i also is expected to increase during summer. However, the area of unvegetated wet fallow lo’i decreased and unvegetated dry fallow lo’i

increased in late summer because some taro farmers did not follow their permit

requirement to leave a harvested field in wet fallow for 30 days. Given the importance of

unvegetated wet fallow lo’i to EWBs, it is highly recommended that the refuge staff

83 ensure that lo’i remain in wet fallow for the full 30 days after harvest is completed as required by permits.

Domestic waterfowl readily consumed seeds of annual plants and insect pests in experimental rice fields of Vietnam (Men et al. 1999). Moorhens, coots, and Koloa may prefer less intensively managed taro lo’i in early growth or mature and medium to dense growth stages for these nutritional benefits. Non-taro emergent vegetation growing in these lo’i may provide additional cover, visual obscurity, and structural support below the taro canopy for overwater nesting moorhens and possibly coots.

ESTIMATING HAWAIIAN COMMON MOORHEN DENSITIES USING CALL-

RESPONSE SURVEYS

An accurate population estimate of the Hawaiian common moorhen is not

available due to the secretive nature of this species and its use of densely vegetated

wetland areas (USFWS 2005). Broadcasts of prerecorded calls to elicit vocalizations are

commonly used to survey secretive marshbirds (Gibbs and Melvin 1993, Ribic et al,

1999, Conway and Gibbs 2005). This technique was successful at improving detection

rates of common moorhen in southwestern Louisiana (Walther and Hohman 1999). Call

response surveys have the potential to provide accurate estimates of the Hawaiian

common moorhen breeding population at HNWR.

Results

The majority of moorhens were detected in taro lo’i during call response surveys

(Table 12). In fact, the mean responses per survey site was greater in taro lo’i ( = 3.1

range = 0-10, n = 552) compared to refuge wetlands ( = 1.8, range = 0-6, n = 96) during

the peak breeding season. Furthermore, there was a significant difference in detections

per survey site between taro lo’i compared to refuge wetlands (two-tailed t-test; t = 4.57; d.f. = 223; p < 0.001).

The number of moorhen detected was similar between peak and off-peak breeding periods (Table 12). However, the number of moorhen responses per survey site in taro

lo’i was significantly different (two-tailed t-test; t = 1.97; d.f. = 235; p < 0.001) between

peak ( = 3.1, range = 0-10, n = 177) and off-peak breeding periods ( = 2.0, range = 0-

9, n = 60). Furthermore, there was only a moderate difference (two-tailed t-test; t = 1.99;

d.f. = 70; p = 0.06) between peak ( = 1.8, range = 0-6, n = 54) and off-peak breeding periods ( = 1.1, range = 0-5, n = 17) in refuge wetlands.

Table 12: Hawaiian common moorhen call-response and visual survey results (subtotal/% of total) by habitat type at Hanalei NWR during the 2004 peak and off-peak breeding season. Taro lo’i Refuge Ditches, pasture Total wetlands and riparian habitat Call response Peak breeding Round 1 106 (85.5%) 12 (9.7%) 6 (4.8%) 124 Round 2 99 (86.8%) 12 (10.5%) 3 (2.6%) 114 Round 3 120 (85.7%) 17 (12.1%) 3 (2.2%) 140 Off-peak breeding Round 1 104 (79.4%) 23 (17.6%) 4 (3.1%) 131 Visual survey Round 3 39.5 (83.2%) 8.0 (16.8%) - 47.5

Moorhens were most likely to be detected during passive or listening portions of the survey and from the edge of the survey plot. Moorhens detected during 5 minutes of passive observation accounted for 46.8% of detections; whereas, moorhens detected during the first and second 3 minutes of listening accounted for 19.5% and 23.8% of detections respectively. Furthermore, 60.2% of moorhens were detected from 40-50 m away.

Repetitive playback of moorhen vocalizations is effective at increasing the number of moorhens detected compared to passive observation alone. Playing 1-minute of taped vocalizations followed by 3 minutes of listening increased the number of

individuals detected by 157.9% over 5 minutes of passive observation alone. Playing

another 1-minute of taped vocalizations followed by 3 minutes of listening increased the number of individuals detected by 208.4% over 5 minutes of passive observation alone.

Consecutive call response surveys are more effective at detecting moorhen than

visual scan surveys. A weekly average of 47.5 moorhens was observed during round 3 of call response surveys (Table 12). A total of 22 of 31 moorhen were observed on 27-28

April 2004, for a detectability index of 71.0%. When the detectability index is applied to

the call response surveys during this period, the estimated moorhen breeding population

is 197 or 414.7% greater than visual scan surveys indicate. The estimated density of

breeding moorhen is 1.6 moorhen/ha in refuge wetlands and 3.6 moorhen/ha in taro lo’i.

Binomial expansion applied to the detectability index indicates two consecutive surveys conducted during the peak breeding season will have a 91.6% certainty of estimating the population size while three consecutive surveys have a 97.6% certainty.

Discussion

Moorhen favor an equal interspersion of water to vegetation where feeding,

nesting, and cover sites are in close proximity (USFWS 1985). Taro lo’i have dense

stands of emergent vegetation that provides thermal cover and visual obscurity from

predators for the majority of the taro agricultural cycle. Taro being harvested, wet fallow

(vegetated and unvegetated), and recently planted lo’i provide moorhens with feeding

habitat that is in close proximity to lo’i with dense overhead cover that serve as escape sites. Mowed dikes that surround lo’i provide additional feeding and loafing sites for

these birds. In comparison, 9.2 ha of refuge wetlands are overgrown with dense stands of

U. mutica that is unsuitable for moorhen nesting or foraging. Call survey results indicate that moorhen are more abundant in taro lo’i than in the refuge wetlands under the conditions during my study. Unfortunately, a wetland infrastructure required for intensive management was not in place throughout my study. Thus, the potential for refuge wetlands to provide habitat for moorhens is unknown.

Detection probability and call response survey effectiveness changes seasonally and optimal survey timing varies by species (Conway and Gibbs 2001). Published estimates of detection rates range from 19% to 100%, but many of these surveys used biased measures of detection probability. Hawaiian common moorhen response rates per survey site varied between peak and off-peak breeding seasons. The number of moorhen that responded also increased during each sampling period of the peak breeding season.

Detection probability was high although the proximity of survey sites and the average

detection distance may bias density estimates.

Movement among tropical resident moorhens can be a response to seasonal

changes in wetland availability. Hawaiian (Nagata 1983, Engilis and Pratt 1993) and

Mariana common moorhen (G. chloropus guami; Takano and Haig 2004) move in

response to droughts and wet periods and readily disperse to seasonal wetlands within islands. Call response surveys suggest that the moorhen population at HNWR is similar during the peak and off-peak periods with shifts in numbers between taro lo’i and refuge

wetlands as feeding habitat become more readily available. A detectability index is not

available to estimate the moorhen population during the off-peak period so the numbers

88 from call response surveys conducted during the peak and off-peak period are not comparable. Moorhen numbers may have increased in refuge wetlands during the nonbreeding season because a large stand of California grass (U. mutica) was mowed and flooded in refuge wetlands before the survey. Removal of this dense vegetation may have increased the effective range of the call survey technique or attracted feeding moorhen.

Hawaiian common moorhen densities in natural wetlands and lotus farms on

O’ahu were 4.6 birds/ha and 17.2 birds/ha respectively (Nagata 1983). These densities are much higher than those observed in refuge wetlands and taro lo’i at HNWR.

However, densities observed on O’ahu may be biased by seasonal differences in moorhen visibility when lotus farms are largely open water habitat prior to planting. Moorhen densities in refuge wetlands are similar to densities of 2.0/ha or less reported in North

American wetlands (Griscom 1933, Bennett and Hendrickson 1939, Bell 1976, Brackney

1979) while densities in taro lo’i are slightly higher.

ENDANGERED WATERBIRD NESTING ECOLOGY

Natality is a key component for the recovery of the Hawai’i’s EWBs. Moorhen

nesting ecology was the focus of early studies at HNWR (Byrd and Zeillemaker 1981); whereas, more recent studies focused on nesting coot and stilt (Asquith and Melgar

1998). These investigations described plants in nesting habitat, but they did not examine habitat characteristics associated with nest-site selection. The objectives of my study were to measure nest habitat characteristics, reproductive success, and to evaluate the relative role of wetland management and taro cultivation in providing suitable habitat for

EWB nesting at HNWR.

Results

Hawaiian common moorhen

A total of 58 Hawaiian common moorhen nests were recorded from mid-January

to late November 2004. April was the peak month with 13 nests. Taro lo’i was the

primary habitat (91.4%); whereas refuge wetlands (3.4%), ditches (3.4%), and dikes

(1.7%) were used occasionally. Taro in early growth to harvesting stages accounted for

77.6% of nests (Figure 29). Moorhen nests were primarily discovered during harvesting

because farmers actively remove the canopy cover in lo’i which makes nests more easily

visible.

Moorhen nests in taro lo’i had medium to high visual obscurity and low to

medium canopy cover. Pooled visual obscurity ranged from 3.9 dm in recently planted

taro to 8.5 dm in medium to dense growth taro (Table 13). Green canopy cover was

between 29.7% and 48.8% for nests found before harvesting and less than 25% following

harvesting (Table 13). Green cover below the canopy level was greater than 10% in

newly planted taro, medium to dense taro, and lo’i being harvested.

Moorhen nest sites in taro lo’i had high plant species richness. A total of 20

species of emergent aquatic vegetation was found at all moorhen nest sites. Each taro

habitat category had less than 2.0 species/nest except recently planted taro lo’i and

vegetated wet fallow because moorhen nests were in patches of non-taro emergent

vegetation (Table 13).

Taro was the dominant vegetation cover at 84.2% of nest sites with a mean height of 60 cm (range = 23-102 cm, n = 48). However, moorhens often built nests in patches of

S. juncoides (17.5%), C. difformis (14.0%), or L. palustris (14.0%) under the taro canopy.

Because plant litter was scarce in taro lo’i (less than 1% at all strata), moorhens searched

for nesting material on dikes and in ditches. The most frequently used nesting materials

were U. mutica (42.1%), taro or Colocasia esculenta (36.2%), Cyperus difformis

(21.1%), and F. littoralis (21.1%).

Microhabitat characteristics were compared between intensive and less

intensively managed taro lo’i. Average number of plant species/vegetated sample plot

and green cover at all strata in less intensively managed taro lo’i was significantly higher

(two-tailed t-test; t = 2.01; d.f. = 53; p < 0.05 for all tests) than in intensively managed

taro lo’i (Table 14). However, no preference was shown in the selection of taro lo’i by

farming intensity (χ2 = 1.593, d.f. = 1, p < 0.21).

The average clutch size was 6.3 eggs (range = 2-11 eggs, n = 56). Egg daily

survival rate (DSR) was 97% and survival until hatching was 57% (n = 783 nesting

days). Once the first egg hatched, the hatching rate was 79% (n = 192 eggs). The proportion of eggs laid that hatched was 46% (n = 352 eggs) and 64% of nests (n = 56

nests) had at least one egg that hatched. The average number of eggs hatched per nest

was 2.9 (range = 0-8 eggs, n = 56). These statistics along with egg metrics (Table 15)

were similar to those measured by Byrd and Zeillemaker (1981). Moorhen chick survival

was only 37% with 60 out of 162 identified as juveniles. Furthermore, the fledging rate

appears to be low (3%) because only 5 juveniles were observed testing their flight

feathers on dikes.

Nest failure was caused by predation (9 nests), abandonment (7 nests), and

flooding (3 nests). The most important nest predators were cattle egrets and/or black-

crowned night-herons (Nycticorax nycticorax; Table 16) which pierced eggs at a single

entry point. Feral cats killed and partially consumed nesting moorhens by accessing nests during drawdowns. Rats also took advantage of drawdowns to predate eggs leaving

scattered shell fragments in the nest. Barn owl (Tyto alba) or short-eared owl (Asio

flammeus) are suspected of killing a nesting moorhen that was found unconsumed. The

necropsy identified evenly spaced dorsal puncture wounds that lead to death by massive

hemorrhaging. A larger number of moorhen nests were predated in less intensively

managed taro lo’i (7 nests) compared to intensively managed taro lo’i (2 nests; Table 17).

Microhabitat characteristics of successful and failed nests were compared.

Canopy cover of failed nests was significantly higher (two-tailed t-test; t = 2.01; d.f. = 53;

p = 0.03) than successful nests (Table 18). However, canopy cover has little impact on

obscuring moorhen nests from predation by cattle egrets and black-crowned night-herons

foraging beneath the upper taro leaves. Canopy cover also has no influence on nest

failure from abandonment or from flooding.

24 22 20 18 s 16 14 12 10

Number ofnest 8 6 4 2 0 planted Recently harvested Lo'i being fallow or mature wet fallow wet Medium to Unvegetated Early growth dense growth Vegetated wet Habitat categories in taro lo'i

Figure 29: Number of Hawaiian common moorhen nests by taro habitat category at Hanalei NWR from 2 February to 18 November 2004 (n = 53).

None None None None None

fallow fallow Vegetated dry Vegetated

None None None None None

dry fallow dry fallow Unvegetated ) at nest microhabitat n

1.8 1.8 3.7 3.7 4.5 13.5 ±

± ± ± SD; 1.1 1.1 = 8) = 8) = 8) = 8) = 5) = 20) = 20) ± n n n n ± n ( ( ( ( ( fallow fallow 3.8% 4.8% 2.4 species 4.3 dm 21.4% Vegetated wet wet Vegetated

0.6 0.6 ±

= 4) = 4) = 1) = 1) = 1) = 1) = 1) = 1) = 1) = 1) i NWR from 2 February to 18 November i NWR from 2 February to 18 November n n n n n 3.5% 0.5% 14.5% ( ( ( ( ( 0.0 species 0.0 species wet fallow wet fallow 4.5 dm Unvegetated

2.1 2.1

16.5 16.5 14.0 14.0 20.9 20.9 ± ± ± ± 2.0 2.0 = 21) = 21) = 21) = 21) = 21) = 21) = 84) = 84) ± = 128) = 128) n n n n n ( ( ( ( ( 1.8 species 1.8 species harvested harvested Lo’i being 5.1 dm 35.8% 11.4% 17.8%

Habitat categories in taro lo’i in taro Habitat categories number of plant species (mean of plant number

2.5 2.5 21.8 21.8 18.3 18.3 10.1 10.1 ± ± ± ±

1.1 1.1 = 13) = 13) = 15) = 15) = 15) = 15) = 15) = 56) = 56) ± n n n n n ( ( ( ( ( 0.5 species 0.5 species 8.5dm 9.8% Medium to 44.8% 15.7% dense growth dense growth

2.2 2.2 5.5 5.5 16.5 5.3 5.3 ± ± ± ± 0.9 0.9 = 10) = 10) = 10) = 10) = 44) = 44) = 10) = 10) ± n n n n n ( ( ( ( ( 6.3% 1.1 species 7.9% or mature or 6.5 dm 29.7% Early growth Early growth iian common moorhen nest sites at Hanale

1.6 1.6 8.5 8.5 18.0 18.0 14.0 14.0 ± ± ± ± = 3) = 3) 3.5 3.5 = 12) = 12) = 3) = 3) = 3) = 3) = 3) = 3) ± n n n n n ( ( ( ( ity, percent green cover, and ( planted Recently Recently 5.5 species 5.5 species 16.0% 3.9 dm 48.8% 26.0%

Green cover (surface) cover Green Pooled visual obscurity Pooled visual 15 cm) (at cover Green (canopy) cover Green species of plant Number (excludes taro) Table 13: Pooled visual obscur sites by taro habitat category of Hawa category habitat sites by taro 2004.

Table 14: Water depth, number of plant species, pooled visual obscurity, and percent green cover (mean ± SD) of Hawaiian common moorhen nests in intensively and less intensively managed taro lo’i at Hanalei NWR from 2 February to 18 November 2004. Intensively Less intensively t p managed taro lo’i managed taro (n = 25) lo’i (n = 30) Water depth 9.0 cm ± 17.3 7.2 cm ± 23.3 2.01 0.14 Distance to dike 6.6 m ± 20.1 7.1 m ± 6.7 2.01 0.58 Number of plant species 0.7 species ± 1.1 2.0 species ± 2.0 2.01 0.01 Pooled visual obscurity 6.1 dm ± 7.1 6.1 dm ±6.7 2.01 0.85 Green cover (surface) 8.8% ± 75.3 18.0% ± 424.2 2.01 0.04 Green cover (at 15 cm) 6.4% ± 24.9 13.0% ± 210.5 2.01 0.02 Green cover (canopy) 28.1% ± 208.5 43.2% ± 396.2 2.01 0.003

Table 15: Hawaiian common moorhen, Hawaiian duck, and Hawaiian stilt egg and nest metrics (mean ± SD) at Hanalei NWR from 23 January to 18 November 2004. Hawaiian common Hawaiian duck (n) Hawaiian stilt (n) moorhen (n) Egg length 45.3 mm ± 2.1 (310) 54.0 mm ± 1.9 (18) 44.6 mm ± 1.8 (151) Egg width 31.9 mm ± 1.5 (310) 39.1 mm ± 2.4 (18) 31.5 mm ± 1.1 (151) Fresh egg weight 24.3 g ± 2.5 (2) - 23.1 g ± 2.3 (22) Nest length 27.5 cm ± 3.7 (57) 30.4 cm ± 5.4 (7) 21.0 cm ± 5.2 (45) Nest width 24.8 cm ± 3.8 (57) 27.7 cm ± 5.2 (7) 19.1 cm ± 4.7 (45) Nest thickness 10.5 cm ± 3.7 (57) 10.6 cm ± 10.3 (7) -

Table 16: Hawaiian common moorhen and Hawaiian stilt nest predators at Hanalei NWR from 2 February to 18 November 2004. Endangered waterbird Hawaiian common moorhen Hawaiian stilt Nest predator Cattle egret and/or black- 4 11 crowned night-heron Feral cat 2 0 Rat 1 0 Owl 1 0 8 11

Table 17: Reasons for Hawaiian common moorhen nest failure in intensively and less intensively managed taro lo’i at Hanalei NWR from 2 February to 18 November 2004. Reason for nest failure Intensively managed Less intensively taro lo’i managed taro lo’i (n = 27) (n = 24) Predation 2 7 Abandonment 3 4 Flooding 2 1

Table 18: Water depth, number of plant species, pooled visual obscurity, and percent green cover (mean ± SD) of successful and failed Hawaiian common moorhen nests at Hanalei NWR from 2 February to 18 November 2004. Successful nests Failed nests t P (n = 35) (n = 21) Water depth 8.0 cm ± 22.6 9.0 cm ± 31.3 2.01 0.47 Distance to dike 6.7 m ± 12.6 6.7 m ± 13.7 2.01 0.97 Number of plant species 2.2 species ± 3.1 2.6 species ± 3.2 2.01 0.48 Pooled visual obscurity 5.9 dm ± 6.3 6.5 dm ± 7.8 2.01 0.12 Green cover (surface) 13.7% ± 286.0 11.2% ± 200.2 2.01 0.58 Green cover (at 15 cm) 9.5% ± 139.8 10.5% ± 173.6 2.01 0.76 Green cover (canopy) 31.2% ± 311.9 42.9% ± 468.3 2.01 0.03

Hawaiian coot

Hawaiian coot broods were observed in taro lo’i (3 broods) and refuge wetlands

(2 broods) from early January to mid-October 2004. Broods observed in taro lo’i were in

close proximity to refuge wetlands. Broods consisted of 3 pairs of chicks and 2 pairs of

juveniles. Fledging success rate could not be determined due to difficulties in resighting

and identifying individual broods. However, half of the chicks were not seen 14-28 days

after first observation.

Hawaiian duck (Koloa)

A total of 7 Koloa nests were discovered in upland areas from mid-January to

mid-May 2004. The majority of nests (5) were discovered on dikes adjacent to refuge

wetlands. Nests also were discovered on the edge of a mowed dike near taro lo’i and in

the garden behind the refuge bunkhouse.

Microhabitat measurements indicate that Koloa nest in tall and dense vegetation

at the refuge. The mean pooled visual obscurity at nest sites was 5.8 dm (range = 0-15

dm, n = 28) for all cardinal directions. Vegetation cover was greater than 50.0% at

canopy height and 33.0% at lower levels (Table 19). Furthermore, visual obscurity (two-

tailed t-test; t = 2.01; d.f. = 46; p = 0.90) and vertical green cover (two-tailed t-test; t =

2.01; d.f. = 46; p > 0.05 for all tests) were not significantly different between Koloa nest

sites and randomly selected sites.

Koloa nest sites had low plant species richness. A total of 8 species of vegetation

were sampled at nest sites with a mean of 2.3 species/nest (range = 1-3 species, n = 7).

Short vegetation such C. diffusa (42.9%) and various forbs (57.2%) were frequently found at nest sites. However, tall grasses such as U. mutica were present at nest sites

more often than randomly selected sites and were also the main nesting material.

The average clutch size was 5.7 eggs (range = 2-7, n = 6) and mean egg

dimensions (Table 14) were within the norm reported by Engilis et al. 2002. Once the

first egg hatched, the hatching rate was 100% (n = 14 eggs). The average number of eggs

hatched per nest was 2.3 (range = 0-7, n = 6) and the proportion of eggs that hatched was

0.41 (n = 34 eggs). Nest success was 57% (n = 7 nests) and nest failure was caused by

abandonment. There was inadequate nesting information to determine mean incubation

period and survival until hatching. Furthermore, fledging success rate could not be

determined due to difficulties in resighting and identifying individual broods.

Table 19: Percent green and litter cover (% mean ± SD) measured at Hawaiian duck (n = 7) and Hawaiian stilt (n = 46) nest microhabitat sites at Hanalei NWR from 23 January to 23 June 2004. Species Green Litter Green Litter Green Litter cover cover cover cover (at cover cover (surface) (surface) (at 15 cm) 15 cm) (canopy) (canopy) Hawaiian 37.3% 16.6% 33.5% 11.3% 55.4% 20.3% duck ± 38.7 ± 25.1 ± 32.3 ± 18.0 ± 38.3 ± 26.6 Hawaiian stilt 6.5% 6.7% 3.2% 0.7% 18.5% 8.5% ± 8.2 ± 10.2 ± 5.1 ± 2.4 ± 14.9 ± 11.5

Hawaiian stilt

A total of 48 Hawaiian stilt nests were monitored from late April to late July 2004

with 35 nests found in the peak nesting month of May. The majority of stilt nests were

discovered in refuge wetlands (87.5%). The highest stilt nest densities were in wetlands

with partial water drawdowns with subunits “C2” and “C1” having 6.1 and 5.8 nests/ha

respectively. In contrast, subunits “A1” and “B1” were rototilled and had nest densities

of 3.9 and 4.3/ha.

Partial water drawdowns creates suitable habitat for nesting stilts more quickly

than rototiller treatments provided that limited residual vegetation is present. Stilt nests

were discovered in wetland subunit “B2” 15 days following a partial water drawdown

compared with 42 days following rototiller treatments in wetland subunit “B1”. Stilts

nested at the edge of tall, residual vegetation in wetland subunit “B2”. In contrast,

vegetation was sparse and low growing in wetland subunit “B1”.

Stilt nests were located primarily on low islands (52.1%), dikes (35.4%), and upland sites (12.5%). Nests were widely spaced with a distance of 100.6 m (range = 8.5-

1,115.0 m, n = 48) between the nearest active nest although stilt nesting was semicolonial at 1 location with 4 nests in a 12.5 m radius. Nesting stilts selected sites with shallower

water compared to the randomly selected sites without nests (two-tailed t-test; t = 2.01;

d.f. = 42; p = 0.003). In fact, nests were on average 1.9 m (range = 0-21 m, n = 46) to the

nearest water.

Microhabitat measurements indicated that stilt nest sites have low visual obscurity

and vegetation cover. The mean pooled visual obscurity for all cardinal directions was

2.0 dm (range = 0-15 dm, n = 184); whereas, green plant and litter cover was less than

20.0% at all strata (Table 19). Nests sites had significantly less plant litter cover at 15 cm

than randomly selected sites (two-tailed t-test; t = 2.01; d.f. = 42; p = 0.04) as this was

likely used as nesting material.

Stilt nest sites had high plant species richness in wetlands. A total of 23 species

of emergent aquatic vegetation was found at stilt nest sites with a mean of 3.8

species/nest (range = 1-8 species, n = 45). The annual sedge, F. littoralis, and perennial

grass, P. distichum, were present at 50.0% and 45.7% of nest sites respectively. Tall

vegetation such as L. octovalvis, U. mutica, and E. crus-galli were present in 54.4%,

43.5%, and 28.3% of nest sites. L. octovalvis was present at nest sites almost twice as

much as it was available at randomly selected sites. When in leaf, the visual obstruction

provided by this tall, woody shrub is important for thermal/escape cover for nesting birds.

This shrub also has low cover at its base that facilitates nest building because there is no

need to remove live and dead vegetation.

The litter at nest sites was most likely to be U. mutica (28.3%), L. octovalvis

(21.7%), and E. crus-galli (19.6%). However, U. mutica was the primary nesting material and was found twice as much as it was available at randomly selected sites. The nest shape was slightly oblong (Table 15) and ranged from bare soil (1 nest was in the

impression of my foot) to fully lined with woven grass.

The average clutch size was 3.3 eggs (range = 1-4 eggs, n = 48) with a mean

incubation period of 25.7 days (range = 24-27 days, n = 6) for nests monitored from the

first egg laid until hatching. The mean egg dimensions and fresh weight (Table 14) were

within the norm for this species (Robinson et al. 1999). DSR of nests was 97% and

survival until hatching was 43% (n = 793 nest days) using the Mayfield method (1975).

After the first egg hatched, the hatching rate was 93% (n = 87 eggs). The proportion of

eggs laid that hatched was 52% (n = 157 eggs) with 48% of nests (n = 48 nests) having at

least one egg that hatched. The average number of eggs hatched per nest was 1.7 (range

= 0-4 eggs, n = 48). The fledging success rate was very low (2.5%) because only 2 of 81

chicks were known to fledge.

Predation, flooding, and abandonment were the main causes of nest failure

accounting for 22.9%, 14.6%, and 14.6% of nests monitored. Cattle egrets and/or black-

crowned night-herons are suspected as nest predators (Table 16) and may have killed one incubating adult. Nests on low islands within the management units had the highest failure rate (64.0%) because they are more at risk from flooding than nests on the foot of

100

levees. All nests surrounded by shallow water were unsuccessful because of flooding or predation. Furthermore, failed nests were located closer to water (two-tailed t-test; t =

2.01; d.f. = 43; p = 0.03) and had lower total green plant cover (two-tailed t-test; t = 2.01;

d.f. = 43; p = 0.04) than successful nests (Table 20). Finally, most abandoned nests were

located on dikes where they are more susceptible to human disturbance.

Table 20: Water depth, number of plant species, pooled visual obscurity, and percent green cover (mean ± SD) of successful and failed Hawaiian stilt nests at Hanalei NWR from 27 April to 23 June 2004. Successful nests Failed nests t p (n = 23) (n = 22) Water depth 0.0 cm ± 0.0 1.1 cm ± 5.2 2.01 0.03 Distance to water 3.3 m ± 39.4 0.6 m ± 1.1 2.01 0.04 Number of plant species 3.8 species ± 3.4 3.7 species ± 3.0 2.01 0.85 Pooled visual obscurity 2.5 dm ± 14.5 1.7 dm ± 6.5 2.01 0.11 Green cover (surface) 9.9% ± 106.3 4.3% ± 18.3 2.01 0.06 Green cover (at 15 cm) 4.3% ± 37.2 2.0% ± 12.9 2.01 0.13 Green cover (canopy) 23.2% ± 203.7 14.2% ± 206.5 2.01 0.04

Endangered waterbird mortality

A total of 45 EWBs (25 Koloa, 9 Hawaiian common moorhens, 6 Hawaiian coots,

2 Hawaiian stilts, and 3 Hawaiian coots/common moorhens) and 4 Hawaiian goose

deaths were noted from 22 January until 2 August 2004. Predation by feral cats was the

most common cause of mortality (Table 21) and increased sharply during March

following the cessation of predator control. Feral cats that prey on birds leave the

remains of wings on dry land which is insufficient information to differentiate moorhens

from coots. Avian botulism deaths were recorded between 3 April and 3 July 2004.

EWB deaths from avian botulism were spread out in the northern section of the refuge in

taro lo’i and refuge wetlands. Dead waterbirds were usually floating with their heads

101

beneath the water’s surface. This disease can potentially impact a host of migratory

waterbirds observed on the refuge (Appendix 24).

Table 21: Causes of EWB mortalities at Hanalei NWR from 22 January to 2 August 2004. Cause of mortality Frequency Percentage of mortalities Cat 15 30.6% Unknown 15 30.6% Avian botulism 10 20.4% Avian predator 4 8.2% Motor vehicle 2 4.1% Exposure/starvation 2 4.1% Malnutrition 1 2.0%

Discussion

Hawaiian common moorhen

The vast majority of nests monitored at the refuge in past (95%; Byrd and

Zeillemaker 1981) studies were in taro lo’i but the condition of refuge wetlands for

moorhen were poorly described. Previous studies have shown that the species of plants

used by moorhens may not be as important as having robust growth of emergents (Karr

and Roth 1971, Weller and Fredrickson 1974). Brackney (1979) reported that moorhen

nest initiation peaked when cattail (Typha spp.) growth rate was greatest and the height of

the cattail was 45-100 cm above the water surface. Taro measured 23-102 cm at

moorhen nest sites. Furthermore, taro’s sturdy stems provide structural support for nests

while it’s broad leaves provide ample thermal cover and visual obscurity from predators.

In fact, nesting moorhen were frequently observed moving their chicks to adjacent lo’i with vegetative cover if taro surrounding their nest was being harvested.

102

Vegetation surrounding moorhen nests in less intensively managed taro lo’i had higher number of plant species and vegetation cover at all strata than intensively managed taro lo’i. Non-taro emergent vegetation provides additional cover, visual obscurity, and structural support below the taro canopy for nesting moorhen. In fact, moorhen nests were found in patches of non-taro emergent vegetation in recently planted and vegetated wet fallow because taro lacks sufficient structural support and thermal cover for nesting in these stages of the agricultural cycle. However, nesting moorhens did not show a preference for less intensively managed taro lo’i based on availability.

Given the estimate of moorhen density (3.6 birds/ha) in taro lo’i using call response surveys, nesting moorhen may have a limited amount of less intensively managed taro lo’i with suitable vegetation structure.

EWB nests have the potential to be located farther from dikes in taro lo’i classified as intensively managed (based on farmer) because they are larger in size than less intensively managed taro lo’i. Nests located farther from dikes provide greater visual obscurity from terrestrial predators. However, the distance of moorhen nests to dikes was not significantly different between intensive and less intensively managed taro lo’i.

Nest success of moorhen at HNWR was 64% and falls within the range of previous studies. For example, reported moorhen nest success rates are 55%-80%

(Cottam and Grazener 1959, Byrd and Zeillemaker 1981, Brackney and Bookhout 1982,

Helm et al. 1987). Predation is the most common cause for nest failure followed by flooding (Cottam and Grazener 1959, Wood 1974, Byrd and Zeillemaker 1981, Brackney

103

and Bookhout 1982, Helm et al. 1987). Similarly, predation and flooding are the main

causes of nest failure at HNWR.

Hawaiian coot

Coot nesting on Kaua’i is highly variable. Part of the variability might be related

to habitat conditions on Ni’ihau. When greater than average rainfall makes more

wetland habitats available for birds on Ni’ihau fewer birds use Kaua’i (Pratt and Bribin

2002, USFWS 2005). However, this theory cannot be substantiated because Ni’ihau is

inaccessible and the relationships between rainfall and nest conditions cannot be verified.

Refuge wetlands and taro lo’i have open water with fairly dense vegetation that

may be used as Hawaiian coot nesting habitat. Although coots were observed brooding,

nests were infrequently encountered at the refuge during the 2004 field season. Nesting

possibly occurred in dense vegetation growing beside or inside ditches.

Hawaiian duck (Koloa)

Most Koloa nests have been found near water within dense shoreline vegetation with no apparent pattern to nest site selection (Engilis et al. 2002). Koloa nests at HNWR

were found in a variety of sites with dense upland vegetation near water except for one

nest found on a mowed dike adjacent to taro lo’i. Dense upland vegetation is abundant

adjacent to most refuge wetlands and ditches. In contrast, dikes adjacent to taro lo’i are

open habitats that provide low thermal cover and lack visual obscurity from nest

predators.

104

Nesting Koloa will abandon eggs if disturbed frequently (Giffin 1983). The

majority of Koloa nests were found near refuge wetlands where human disturbance is

lower than in taro lo’i. Observer disturbance may have caused abandonment on refuge

wetlands for at least one nest.

Hawaiian stilt

Wetland management plays an important role in Hawaiian stilt nesting at HNWR.

Refuge staff can create suitable habitat for breeding stilts in a timely manner. Of the 48 nest found on the refuge, 42 were at sites specifically managed for stilt nesting. This response by nesting stilts indicates the potential for intensive management on refuge wetlands. For instance, partial water drawdowns in wetland subunits “C1” and “C2” during mid-April 2004 created mudflats surrounding residual vegetation suitable for nesting stilts. Within 19 days of decreasing water levels, a total of 7 nests were discovered at unflooded sites within these subunits. In contrast, taro farmers control water levels and vegetation cover in taro lo’i and nesting started 1 month later than in refuge wetlands. Stilt nests on taro dikes have less risk from disturbance during harvesting or from flooding for taro management. However, stilt nests are less secure because terrestrial predators readily access stilt nests on taro dikes where there is little visual obscurity and human disturbance is greater from taro management activities.

Stilts nested on mudflats in managed wetlands where sparse, low growing vegetation was present following soil disturbance or on the edge of tall residual vegetation following partial water drawdowns. Stilts nested predominantly in sparse, low

105 growing vegetation that germinated following rototiller treatments at Kealia Pond NWR,

Hawaii (Rader 2005). Stilt nests were surrounded by a mean of 56% vegetative cover at

Bolsa Chica, California (James 1995). Wetlands dominated by open water habitat were chosen for partial drawdown at HNWR. When wetlands were partially drawndown, there was a greater interspersion of water to vegetation (unpublished data). Thus, nest densities were higher on sites with partial drawdowns.

Nest success of Hawaiian stilts (48%) at HNWR was lower than previous studies.

Predation and flooding were the main causes of nest failure. Nest success of stilts ranged from 54-61% on other Hawaiian Islands (Chang 1990, Rader 2005). Predation was the major cause of nest failure for two studies at James Campbell NWR, accounting for 26%

(Coleman 1981) and 25% (Chang 1990) of losses respectively. Flooding contributed to

25% of nest losses at Kealia Pond NWR (Rader 2005). Nest success was highest on islands that protect nests from ground predators and flooding (Robinson et al. 1999).

However, nest failure at HNWR was highest on low islands because stilt nests were located close to water where sudden and intense rainfall quickly flooded nests.

Endangered waterbird recruitment

Predation, abandonment, and flooding were the main causes of nest failure for

EWBs at HNWR. Feral cats and rats predated stilt nests on dikes; whereas, water drawdowns provided access to moorhen and stilt nests located in taro lo’i and refuge wetlands. Cattle egrets, black-crowned night-herons, and owls predated moorhen and stilt nests regardless of the presence of water. Koloa may abandon nests with frequent

106

human disturbance (Giffin 1983); whereas, moorhen nest abandonment from disturbance

appeared to decrease later during the incubation period. Because limited infrastructure exists to control water levels in taro lo’i and refuge wetlands, sudden and intense rainfall can quickly overtop nests of ground and over-water nesting species.

My estimation of chick survival was more reliable for stilts compared with moorhens because stilts are more frequently observed in open habitats. However, moorhen and stilt chick survival at HNWR may be low due to predation. Feral cats, rats, dogs, cattle egrets, black-crowned night-herons, barn owls, and short-eared owls are possible predators of stilt chicks at the refuge. Furthermore, bullfrogs are present on the refuge and were identified as a key predator of juvenile stilts in a radio-telemetry study conducted at James Campbell NWR on O’ahu (Englund 2004).

Introduced tilapia (Seratherodon melanotheron) may degrade feeding habitats by depleting the invertebrate prey base needed by waterbird chicks (USFWS 2005). A fish exclosure experiment at Kealia Pond NWR demonstrated that tilapia could drastically decrease the abundance of chironomids and significantly decrease the abundance of corixids (N. L. Wirwa, pers. comm.). An invertebrate field study in progress at HNWR will evaluate availability of these food resources for growing chicks. However, egg dimensions and mass are within the norm for moorhen (Byrd and Zeillemaker 1981) and stilts (Robinson et al. 1999) suggesting that egg-laying waterbirds have adequate food resources at the refuge (Cody 1971, Batt and Prince 1979).

Adult EWB mortalities caused by feral cats and avian botulism can be reduced with careful management. An effective predator control program has the potential to

107

reduce feral cat populations on the refuge but such programs are expensive and their

effectiveness depends on the amount of personnel time available as well as the trapping

skills of the individual. The predator control program on the refuge was discontinued

beginning in March 2004, but what effect this change in management had on my results is unknown. Avian botulism has reappeared annually on the refuge in recent years and can affect the health of all native and migratory waterbirds. Refuge staff should continue monitoring outbreaks of avian botulism and quickly remove dead and dying birds that contribute to the spread of this disease (USFWS 2005). Furthermore, refuge staff should flush affected wetlands with fresh water to remove the toxins that can cause mortality.

108

MANAGEMENT RECOMMENDATIONS AND RESEARCH NEEDS

Management recommendations

Taro cultivation

The structural condition of taro lo’i was classed in 8 categories based on the cover

provided in combination with the hydrologic regime. As taro passes through these

stages, the length of time in each stage is dependent on the variety of taro, soil texture

and chemistry, water management, the rate and degree of fertilizer application, and

intensity for removal of non-taro emergent vegetation. At anytime, approximately 50-

60% of lo’i on the refuge are in early growth or mature and medium to dense growth forms. These dense taro growth stages provide important thermal/escape cover for nesting moorhen and possibly coot year-round.

Among the 8 taro growth categories, all four EWBs preferred taro habitat categories during harvesting or the wet fallow period (vegetated and unvegetated). The availability of water during the entire year is important for creating habitat conditions suitable for EWBs. Thus, the water distribution system for taro on the refuge must provide a sufficient volume of water to flood fallow fields that will create the wet conditions preferred by waterbirds. This is especially true during the summer months when water discharge from the Hanalei River is the least.

Moist-soil plants such as Cyperus difformis, C. polystachyos, Echinochloa crus- galli, Eleocharis obtusa, Fimbristylis dichotoma, F. littoralis, Kyllinga brevifolia, and

Schoenoplectus juncoides were common in vegetated dry fallow taro lo’i. Fields

109

characterized by these plants also had high densities of moorhens, coots, and Koloa.

Flooding of vegetated dry fallow fields during rainfall events attracted high densities of these EWBs. Thus, fallow taro fields have the potential to attract EWBs by mimicing

sudden flooding of moist-soil vegetation using irrigation water. The response from F. littoralis in managed wetlands indicated that the time period required to optimize benefits for EWBs was 94 days. However, more information is required to determine the amount

of time for Cyperus difformis, C. polystachyos, Echinochloa crus-galli, Eleocharis

obtusa, Fimbristylis dichotoma, K. brevifolia, and S. juncoides to develop seed heads

following rototiller treatments on fine and coarse-textured soils. Maintaining wet fallow

for the 30 day time period required by existing special use permits for taro cultivation on

the refuge is an important guideline for EWB management at HNWR.

Aquatic invertebrates are an important food source for nesting waterbirds.

Invertebrate diversity and biomass is dependent on algae and the structure of emergent

plant communities (Fredrickson 1996, Davis and Smith 1998, Anderson and Smith 1999).

Application of nitrogen fertilizer encourages the growth of algae, which is an important

component of the invertebrate food chain. However, intensive removal of non-taro

emergent vegetation during the first 4-6 months of taro growth decreases the vegetation

structure available to invertebrates as well as the structural support required for moorhen

nests. Annual plants such as Cyperus polystachyos, C. difformis, Eleocharis obtusa, F.

littoralis, and S. juncoides also provide forage for moorhen, coot, and Koloa. Thus, less

intensive taro cultivation practices have the potential to provide food resources for all

EWB species and suitable vertical structure for moorhen nesting. Incorporation of

110

findings from the invertebrate study in progress at HNWR will provide insight into the more specific relationships of the invertebrate resource in relation to avian use and taro

stage.

Wetland management

The HNWR staff plans to reconfigure wetland subunits “B3-16” into fewer subunits that account for differences in soil properties. Much of this reconfiguration was completed by the end of 2006. This reconfiguration allows for individual wetland management strategies within most subunits and is aligned with soil texture differences.

This realignment is important because germination and plant growth differ according to textural differences. Enlarging the units will decrease the edge effect and provide a buffer for human disturbance or predator access. Unit size is important because stilt and other nests can be located further from dikes, where they likely will be less at risk from terrestrial predators during drawdowns.

Other refuge wetlands that can be improved for EWBs include the “Rice Mill” and “DU” ponds. In the “Rice Mill Ponds”, identifying different soil textures should be used to reconfigure these wetlands to provide the highest quality habitat for EWBs.

Within “DU Pond”, the north-south elevational differences represent a constraint because water levels must be high in order to provide surface flooding on the southern portion of this wetland. When ideal conditions are produced on the southern shallower portion of the “DU Unit”, the northern section of deepwater habitat lacks conditions for the growth of emergent aquatic vegetation and is too deep for optimal foraging by EWBs. By

111 decreasing the elevational differences, more of the pond can provide optimal habitat for the production of food as well as desirable foraging depths. Furthermore, the small islands should be reconfigured to encourage the growth of wetland vegetation rather than the invasive plants that now readily infest these sites because of their high elevation.

The tillage treatment of refuge wetlands created mudflats for germination of early successional wetland plants and topographic conditions suitable for nesting stilts.

Manipulation of water levels after soil disturbance attracted feeding EWBs and promoted seed germination of wetland plants. Because different seeds germinate in response to a wide suite of environmental conditions (e.g., soil temperature, soil moisture, soil textures, salinities, pH, and other abiotic factors), an effort should be made to create conditions required for desirable plant species to germinate.

To meet life-history needs of EWBs, the managed wetlands must provide food and cover for roosting and nesting. For example, F. littoralis provides seeds utilized by

Koloa. Because F. littoralis grows on fine and coarse-textured soils, management of this plant with rototiller treatments can be implemented throughout the area encompassed by refuge wetlands. In contrast, P. distichum grows primarily on coarse-textured soils that are continuously flooded and provides invertebrate substrates and seeds but it does not provide structure suitable for nesting coot and moorhen. Further information is needed to determine the amount of time for Cyperus difformis, C. polystachyos, Eleocharis obtusa,

F. dichotoma, K. brevifolia, and S. juncoides to develop seed heads following rototiller treatments on fine and coarse-textured soils. Other plant species such as Echinochloa crus-galli provide tall and more robust dense vegetation of value for nesting habitat for

112

coot, moorhen, and Koloa. For example, Echinochloa crus-galli provided high visual

obscurity for stilt and moorhen nests in refuge wetlands. This annual plant is easily managed on coarse-textured soils and its seeds are an excellent source of carbohydrates and minerals for Koloa (Fredrickson and Taylor 1982). Furthermore, tall and dense

vegetation on edges of wetlands and ditches provides nesting habitat for coots and Koloa.

Such habitat also benefits the Hawaiian goose, which is another endangered bird found

on the refuge. Thus, management strategies for individual wetlands must have well

developed habitat objectives that consider topography, soils, and current vegetation

conditions in order to provide feeding or nesting cover for specific EWBs and other

wetland-dependent wildlife.

Dike management

Grass-covered dikes surrounding refuge wetlands and taro lo’i should be managed

as habitat for EWBs but the dikes on each habitat have different management potential.

The majority of foraging and loafing EWBs were observed on mowed dikes surrounding taro because frequent mowing encourages vigorous plant growth and enhances visibility for predator detection. In contrast, the majority of dikes surrounding refuge wetlands was infrequently mowed and was used as cover for nesting Koloa. The edge of dikes surrounding refuge wetlands should be left unmowed to provide food and cover for nesting Koloa, as well as nesting material for coots and moorhens. Furthermore, grass- covered dikes separating taro lo’i and refuge wetlands should be left unmowed to provide the same benefits for nesting EWBs.

113

Endangered waterbird recruitment

A key to protecting and enhancing EWB populations at HNWR is improving the nesting and fledging success rates. Nest predators at the refuge during 2004 were cattle egrets, black-crowned night-herons, feral cats, rats, and owls. Nesting success also was affected by abandonment resulting from human disturbance and the effects on nesting may be influenced by the experience of nesting females. Although there are no public uses on HNWR that disturb nesting EWBs in refuge wetlands, taro lo’i are constantly being disturbed from taro cultivation practices. The behavioral effects of disturbance vary with each species. For example, nesting moorhens have been observed both abandoning or returning to nests in taro lo’i where taro farmers were harvesting or weeding nearby. In contrast, nesting Koloa abandon nests when disturbance is frequent

(Giffin 1983). Therefore, once Koloa nests have been located disturbance should be limited.

Removing EWB predators, monitoring and preventing their presence, or controlling their distribution by the presence of surface water has potential benefits to

EWBs on HNWR. Potentially refuge staff might haze or kill cattle egrets and black- crowned night-herons when they are foraging near EWB nest sites. Water level fluctuations also are an important factor in nesting success within taro lo’i and refuge wetlands. For example, moorhens abandon marshes left dry by sudden drawdowns

(Brackney 1979, Helm 1982). Therefore, drawdowns should not be permitted in taro lo’i when a moorhen nest is active because terrestrial predators (e.g., feral cats and rats) have better access and the nest may collapse when water levels decrease too rapidly. Because

114

bullfrogs are a known predator of stilt chicks on O’ahu (Silbernagle and Eijzenga, pers. comm.) monitoring their presence is important. With the new infrastructure in refuge wetlands, the reduction in deep water habitats within the management units will likely reduce bullfrog numbers within units with silt nesting.

A moorhen was observed increasing the height of its nest by over 25 cm to avoid

destruction by flooding when water levels increased gradually in a taro lo’i. However,

the risk of flooding on nesting EWBs within taro lo’i and refuge wetlands is

unpredictable because rainfall at the refuge can be sudden and intense which overtops

stilt nests quickly or increases water levels faster than over water nesting species like

moorhen and coot can add materials to their nests. Wetlands should be configured to

provide an equal interspersion of water to mudflats that decreases nest loss from flooding.

In addition, water control structures need to be modified to control rapidly increasing

water in refuge wetlands to protect stilts nests susceptible to flooding during intense

rainfall events. Furthermore, taro permitees must carefully monitor water levels in taro

lo’i with known moorhen nests to avoid accidental flooding when drain outlets become blocked by plant debris.

Avian botulism is the most prevalent disease affecting EWBs at HNWR. This disease reappears annually and can affect all native and migratory waterbirds. Tracking of the location and timing of avian botulism outbreaks might reveal patterns that could be used to avoid environmental conditions that lead to outbreaks (USFWS 2005). The refuge staff should continue monitoring outbreaks of avian botulism and quickly remove dead and dying birds that contribute to the spread of this fatal disease. Furthermore,

115

refuge staff should flush affected wetlands with fresh water to remove the toxins that can

cause mortality.

Research needs

Population estimates and the distribution of species are needed to assess the status

of EWB species at HNWR but the time required for these estimates must match the time

that can be assigned for monitoring populations. For example, surveys might be

scheduled for different purposes linked with habitat objectives (e.g., nesting, movement

related to weather or season, and life cycle events). Based on my experience, I suggest

that total counts take place beginning at sunrise at a carefully selected interval related to

avian life history events and/or plant conditions. Total counts are possible from survey

sites on the bluffs surrounding the refuge (see Figure 17). A spotting scope with an 80

mm objective lens and 20-60X magnification provides the best view of EWBs in distant

impoundments. The best potential to generate information on the number and

distribution of the waterbirds across all refuge habitats without observer disruptions requires observations from a distance as well as visibility across the different structural conditions in taro lo’i and refuge wetlands. More detailed information on the distribution and use of habitats are not possible from a distance. Conducting ground surveys on randomly selected survey sites with different habitat conditions or during different life cycle events would provide much better information on species response to habitat conditions and result in a visibility index. Refining a visibility index for each EWB species within each taro habitat category and refuge wetland cover class would enhance

116 the value of the visual survey. A visibility index must also take into account time of day, distance to the impoundment, and level of human disturbance.

Call response surveys give accurate estimates of moorhen during the peak breeding months of April and May but must include the following factors in the sampling protocol—start 30 minutes before sunrise until 4 hours after sunrise; survey sites should be located at least 100 m apart on dikes to avoid double-counting; only moorhen within

50 m should be recorded unless the response is not from taro lo’i or refuge wetlands; use recordings of the Hawaiian subspecies of common moorhen; use audio devices which produce >80 decibels of sound from a 1 m distance (Gibbs and Melvin 1993); at least two experienced surveyors should conduct the surveys from different parts of the refuge at the same time; the order and time at which sites are visited should be randomized to control for bias; and survey sites should be monitored 2-3 times separated by 7 days.

Seeds provide resources important for the nutritional needs of waterbirds that assure a body condition for successful molting and reproduction. Estimating seed production after various treatments and time periods provides insights into the success of wetland manipulations to provide these resources. Regression equations to estimate seed production are available for E. crus-galli (Laubhan 1992, Laubhan and Fredrickson

1992). Regression equations also are available for two species of Cyperus (C. esculentus and C. erythrorhizos; Laubhan 1992), thus these equations might be tested with Hawaiian members of the genus Cyperus with similar growth characteristics. Nevertheless, developing regressions for early successional native and naturalized Hawaiian plants such as F. littoralis has value in estimating foods available for EWB use.

117

Radio-telemetry studies are needed to provide a better understanding of seasonal fluctuations in the abundance of coots, stilts, and Koloa in Hawai’i. Seasonal EWB surveys from other islands (especially from Ni’ihau) in combination with weather data will provide empirical evidence for the theory of intra and interisland movement of

EWBs based on differences in flooding conditions in wetlands between the wet and dry seasons. Furthermore, radio-telemetry studies can provide insights into EWB dispersal, philopatry, and survival on the refuge.

Summary

This study has provided information on the relative contributions of wetland management and taro cultivation in providing suitable habitat for Hawaiian EWBs in a year-round growing season. Each stage of the taro agricultural cycle provides different habitat conditions that contribute to the life-history requirements (e.g., foraging, loafing, and nesting) for EWBs. The availability of these taro lo’i habitat conditions is influenced by the stage and management intensity of taro in combination with variation in the market demand for poi during the year. In contrast, managed wetlands on the refuge can be manipulated specifically to provide foraging and nesting habitats to meet life cycle requirements of each EWB species as needed. During my study, EWB response to habitat conditions following soil disturbance illustrate the potential of wetland management for EWBs at the refuge. However, the infrastructure for wetland management (water distribution system) on HNWR was inadequate to consistently provide a diversity of habitat conditions for the four EWBs during my study.

118

Consequently, it is highly recommended that refuge wetland units be reconfigured into larger units based on soil texture and a water distribution system be developed to allow independent water supply to create habitat conditions suitable for all EWB life-history stages.

119

LITERATURE CITED

Alberico, J. A. R. 1995. Floating eggs to estimate incubation stage does not affect hatchability. Wildlife Society Bulletin 23:212-216

Altman, J. 1974. Observational study of behavior: sampling methods. Behavior 49:227- 267.

Anderson, J. T. and L. M. Smith. 1999. Carrying capacity and diel use of managed playa wetlands by nonbreeding waterbirds. Wildlife Society Bulletin 27:281-291.

Asquith, A. and C. Melgar. 1998. Waterbird report for Hanalei and Huleia NWRs: 1996 & 1997. Prepared for Kauai National Wildlife Refuge Complex, 18pp.

Bannor, B. K. and E. Kiviat. 2002. Common Moorhen (Gallinula chloropus), in The Birds of North America, No.685. (A. Poole and F. Gill, eds.). The Birds of North America, Inc., Philadelphia, PA, 28pp.

Batt, B. D. and H. H. Prince. 1979. Laying dates, clutch size and egg weight of captive mallards. Condor 81:35-41.

Bell, G. R. 1976. Ecological observations of Common (Gallinula chloropus) and Purple gallinules (Porphyrula martinica) on Lacassine National Wildlife Refuge, Cameron Parish, Louisiana, M.S. Thesis, University of Southwestern Louisiana, Lafayette, LA.

Bennett, L. J. and G. O. Hendrickson. 1939. Adaptability of birds to changed environment. Auk 56:32-37.

Brackney, A. W. 1979. Population ecology of common gallinules in southwestern Lake Erie marshes. M.S. Thesis, Ohio State University, Columbus, OH, 69pp.

Brackney, A. W. and T. A. Bookhout. 1982. Population ecology of Common Gallinules in southwestern Lake Erie marshes. Ohio Journal of Science 82:229-237.

Brenowitz, E. A. 1981. The effect of stimulus presentation sequence on the response of Red-winged Blackbirds in playback studies. Auk 98:355-360.

Brisbin, I. H. Jr. and T. B. Mowbray. 2002. American Coot (Fulica alai), in The Birds of North America, No.697. (A. Poole and F. Gill, eds.). The Birds of North America, Inc., Philadelphia, PA, 44pp.

120

Broshears, R. E. 1979. The influence of trophic interaction on the distribution and abundance of selected aquatic species in a Hawaiian taro ponds ecosystem, M.S. thesis, University of Hawaii, Honolulu, HI, 63pp.

Byrd, G. V. and C. F. Zeillemaker. 1981. Ecology of nesting Hawaiian Common Gallinules at Hanalei, Hawaii. Western Birds 12:105-116.

Byrd, G. V., R. A. Coleman, R. J. Shallenberger, and C. S. Arume. 1985. Notes on the breeding biology of the Hawaiian Race of the American Coot. ‘Elepaio 45:57-63.

Chabreck, R. H. 1968. The relation of cattle and cattle grazing to marsh wildlife and plants in Louisiana. Proceedings, Annual Conference Southeastern Association of Game and Fish Commissioners 22:55-58.

Chang, P. R. 1990. Strategies for managing endangered waterbirds on Hawaiian National Wildlife Refuges. M.S. Thesis, University of Massachusetts, Amherst, MA.

Chin, D. V, T. T. N. Son, L. C. Kiet, K. Itoh, H. Hiraoka, and H. Kobyashi. 2001. Lowland rice weeds in Vietnam. Proceedings of the 2001 Annual Workshop of JIRCAS. September 12-21, 2001. 4pp. (2006, February 18).

Cody, M. L. 1971. Ecological aspects of reproduction. Pages 461-512 in Avian Biology, Vol. 1 (D. S. Farner and J. R. King, eds.). Academic Press, New York, NY, 343pp.

Coleman, R. A. 1979. Wetlands, waterbirds, and Hawaii Audubon. ‘Elepaio 40:73-75.

Coleman, R. A. 1981. The reproductive biology of the Hawaiian subspecies of the Black-necked Stilt, Himantopus mexicanus knudseni. Ph.D. diss., Pennsylvania State University, University Park, PA, 106pp.

Conway, C. J. and J. P. Gibbs. 2001. Factors influencing detection probability and the benefits of call broadcast surveys for monitoring marsh birds. Final Report, U.S. Geological Survey, Patuxtent Wildlife Research Center, Laurel, MD.

Conway, C .J. and J. P. Gibbs. 2005. Effectiveness of call-broadcast surveys for monitoring marsh birds. Auk 122:26-35.

Cottam, C. and W. C. Grazener. 1959. Late nesting of waterbirds in south Texas. Transcripts of the North American Wildlife Conference 24:383-395.

121

Daubenmire, R. 1959. A canopy-coverage method of vegetational analysis. Northwest Science 33:43-64.

Davis, C. A. and L. M. Smith. 1998. Ecology and management of migrant shorebirds in the Playa Lakes Region of Texas. Wildlife Monograph 140:1-45.

De la Pena, R. S. 1972. Effects of nitrogen fertilization on the growth, composition and yield of upland lowland taro (Colocasia esculenta). Experimental Agriculture 8:187-194.

De la Pena, R. S. 1977. Effects of plant density on the yields and yield components of upland and lowland taro, Colocasia esculenta (L.) Schott. Technical paper 174:131-138.

De la Pena, R. S. 1983. Agronomy. Pages 167-179 in J.A. Wang (ed.). Taro, a review of Colocasia esculenta and its potentials. University of Hawaii Press, Honolulu, HI.

DesRochers, B. A. and C. D. Ankney. 1986. Effect of brood size and age on the feeding behavior of adult and juvenile American Coots (Fulica americana). Canadian Journal of Zoology 64:1200-1406.

Drilling, N, R. Titman, and F. McKinney. 2002. Mallard (Anas platyrhynchos) in The Birds of North America, No.58. (A. Poole and F. Gill, eds.). The Birds of North America, Inc., Philadelphia, PA, 42pp.

Driver, E. A. 1988. Diet and behaviour of young American Coots. Wildfowl 39:34-42.

Emlen, J. T. 1971. Population densities of birds derived from transect counts. Auk 88:323-342.

Emlen, J. T. 1977. Estimating breeding season bird densities from transect counts. Auk 94:455-468.

Endangered Species Information System. 1996a, March 14. (Draft) – Taxonomy. Species: Common moorhen, Hawaiian. Species Id ESIS101014. (2003, February 13).

Endangered Species Information System. 1996b, March 14. (Draft) – Taxonomy. Species: Coot, Hawaiian. Species Id ESIS101030.

1 (2003, February 13).

122

Endangered Species Information System. 1996c, March 14. (Draft) – Taxonomy. Species: Stilt, Hawaiian. Species Id ESIS101013. (2003, February 13).

Engilis, A. and T. K. Pratt. 1993. Status and population trends of Hawaii’s native waterbirds, 1977-1987. Wilson Bulletin 105:142-158.

Engilis, A., Jr. and K. Y. J. Uyehara and J. G. Giffin. 2002. Hawaiian Duck (Anas wylvilliana) in The Birds of North America, No.694. (A. Poole and F. Gill, eds.). The Birds of North America, Inc., Philadelphia, PA, 20pp.

Englund, R. A. 2004. Identifying key predators of Hawaiian stilt (Himantopus mexicanus knudseni) chicks. Page 26 in The 2004 Hawaii Conservation Conference: Hawaii’s Invasive Species Challenges. Hilton Hawaiian Village June 29-30, 2004. Hawaii Conservation Alliance. Honolulu, HI.

Falls, J. B., T. E. Dickinson, and J. R. Krebs. 1990. Contrast between successive songs affecting the response of eastern meadowlarks to playback. Animal Behaviour 39:717-728.

Fredrickson, L. H. 1971. Common gallinule breeding biology and development. Auk 88:914-917.

Fredrickson, L. H. 1996. Moist-soil management, 30 years of field experimentation. International Waterfowl Symposium 7:168-177.

Fredrickson, L. H. and T. S. Taylor. 1982. Management of seasonally flooded impoundments for wildlife. U.S. Fish and Wildlife Service Resource Publication 148.

Gibbs, J. P. and S. M. Melvin. 1993. Call-response surveys for monitoring breeding waterbirds. Journal of Wildlife Management 57:27-34.

Giffin, J. G. 1983. Abundance, and distribution of Koloa on the Island of Hawai’i: movements, survival, reproductive success and habitat of Koloa on the Island of Hawai’i. Pittman-Robertson report (W-18-R-7-19-8, R-III-H). Hawai’i Division of Forestry and Wildlife, Honolulu, HI.

Griscom, L. 1933. The birds of Dutchess County, New York. Transactions of the Linnean Society of New York. Volume 3.

Hardy, D. E. 1960. Insects of Hawaii Vol. 10 Diptera: Nematocera-Brachycera. University of Hawaii Press, Honolulu, HI.

123

Helm, R. N. 1982. Chronological nesting study of common and purple gallinules in the marshlands and rice fields of southwest Louisiana. M.S. Thesis, Louisiana State University, Baton Rouge, LA, 114pp.

Helm, R. N., D. N. Pashley, and P. J. Zwank. 1987. Notes on the nesting of the Common Moorhen and Purple Gallinule in southwestern Louisiana. Journal of Field Ornithology 58:55-61.

Higgins, K. F., J. L. Oldemeyer, K. J. Jenkins, G. K. Clambey, and R. F. Harlow. 1996. Vegetation sampling and measurement. Pages 567-591 in T. A. Bookhout (ed.). Research and management techniques for wildlife and habitats. Fifth ed., rev. The Wildlife Society, Bethesda, MD.

Hue, N. V, F. Guo, G. Zhang, R. S. Yost, and S. C. Miyaska. 1997. Reactions of Copper Sulfate with Wetland-Taro soils in Hawaii. Communications in Soil Science and Plant Analysis 28:849-862.

James, R.A., Jr. 1995. Sex roles in parental care of the Black-necked Stilt (Himantopus mexicanus) in southern California. M.S. Thesis, University of California, Los Angeles, CA.

Johnson, D. H. 1980. The comparison of usage and availability measurements for evaluating resource preference. Ecology 61:65-71.

Kadlec, J. A. and L. M. Smith. 1989. The Great Basin marshes. Pages 451-474 in Habitat management for migrating and wintering waterfowl in North America. (L. M. Smith, R. L. Pederson, and R. M. Kaminski, eds.). Texas Tech University Press, Lubbock, TX, 574pp.

Karr, J. R. and R. R. Roth. 1971. Vegetation structure and avian diversity in several new world areas. American Naturalist 105:423-435.

Kent, R. J. and D. E. Johnson. 2001. Influence of flood depth and duration on growth of lowland rice weeds, Côte d’Ivoire. Crop protection 20:691-694.

Laubhan, M. K. 1992. A technique for estimating seed production of common moist-soil plants. Waterfowl Management Handbook. Fish and Wildlife Service Leaflet 13.4.5. U.S. Fish and Wildlife Service, Washington D.C., 8pp.

Laubhan M. K. and L. H. Fredrickson. 1992. Estimating seed production of common plants in seasonally flooded wetlands. Journal of Wildlife Management 56:329– 337.

124

Liou, T. D. 1984. Effect of plant density on the yield of taro in paddy field. Journal of Agricultural Research of China 33:38-43.

Lu, H. U., C. T. Lu, L. F. Chan, and M. L. Wei. 2001. Seasonal variation in linear increase of taro harvest index explained by growing degree days. Agronomy Journal 93:1136-1141.

Manrique, L. A. 1994. Nitrogen requirements of taro. Journal of Plant Nutrition 17:1429-1441.

Mayfield, H. 1975. Suggestions for calculating nest success. Wilson Bulletin 87:456- 466.

McCracken, J. D. 1987. American coot. Pages 158-159 in Atlas of the breeding birds of Ontario (M. D. Cadman, P. F. L. Eagles, and F. M. Helleiner, eds.) University of Waterloo Press, Waterloo, ON.

Men, B. X., T. K. Tinh, T. R. Preston, R. B. Ogle, and J. E. Lindberg. 1999. Use of local ducklings to control insect pests and weeds in the growing rice field. Livestock Research for Local Development 11:2

Mitsch W. J. and J. G. Gosselink. 1993. Wetlands. 2nd Edition. Van Nostrand Rienhold Publications. New York, NY, 722pp.

Mulholland, R. and H. F. Percival. 1982. Food habits of the Common Moorhen and Purple Gallinule in north-central Florida. Proceedings of the Annual Conference of the Southeastern Association of Fish and Wildlife Agencies 36:527-536.

Murkin, H. R. 1989. The basis for food chains in prairie wetlands. Pages 316-38 in A. G. Van der Valk (ed.). Northern Prairie Wetlands. Iowa State University Press, Ames, IA.

Nagata, S. E. 1983. Status of the Hawaiian Gallinule on lotus farms and a marsh on O’ahu, Hawai’i. M.S. thesis, Colorado State University, CO, 87pp.

National Agriculture Statistics Service. 2005. Hawaii Taro. United States Department of Agriculture, 4pp.

Olson, S. L. and H. F. James. 1992. Fossil birds from the Hawaiian Islands: evidence for wholesale extinction by man before western contact. Science 217:633-635.

125

Paveglio, F., A. Engilis, J. Kauffeld, M. Silbernagle, G. Zalm, and R. Shallenberger. 1999. Hawaiian Wetland Refuge Habitat Management Review, April 7-16, 1999. U.S. Fish and Wildlife Service, Refuges and Wildlife, Region 1, Portland, OR, 24pp.

Pons, T. L. and H. F. J. M. Schröder. 1986. Significance of temperature fluctuation and oxygen concentration for germination of the rice field weeds Fimbristylis littoralis and Scirpus juncoides. Oecologia 68: 315-319.

Pratt, H. D. and I. L. Bribin Jr.. 2002. Hawaiian Coot (Fulica alai), in The Birds of North America, No.697. (A. Poole and F. Gill, eds.). The Birds of North America, Inc., Philadelphia, PA, 44pp.

Rader, J. A. 2005. Response of vegetation and endangered waterbirds to habitat management techniques at Kealia Pond National Wildlife Refuge. M.S. Thesis, South Dakota State University, Brookings, SD, 93pp.

Reed, J. M., M. D. Silbernagle, K. Evans, A. Engilis Jr., and L. W. Oring. 1998. Subadult movement patterns of the endangered Hawaiian stilt (Himantopus mexicanus knudseni). Auk 115:791-797.

Reinecke, K. H., R. M. Kaminski, D. J. Moorhead, J. D. Hodges, and J. R. Nassar. 1989. Mississippi Alluvial Valley. Pages 203-247 in Habitat management for migrating and wintering waterfowl in North America (L. M. Smith, R. L. Pederson, and R. M. Kaminski, eds.). Texas Tech University Press, Lubbock, TX, 574 pp.

Ribic, C. A., S. J. Lewis, S. Melvin, J. Bart, and B. Peterjohn. 1999. Workshop results. Pages 5-18 in Proceedings of the marsh bird monitoring workshop, Laurel Maryland, April 1999. U.S. Fish & Wildlife Service and U.S. Geological Survey, Laurel, MD.

Robel, R. J., J. N. Briggs, A. D. Dayton and L. C. Hulbert. 1970. Relationships between visual obstruction measurements and weight of grassland vegetation. Journal of Range Management 23:296-297.

Robinson, J. A., J. M. Reed, J. P. Skorupa, and L. W. Oring. 1999. Black-necked stilt (Himantopus mexicanus), in The Birds of North America, No. 449 (A. Poole and F. Gill, eds.). The Birds of North America, Inc., Philadelphia, PA, 32pp.

Schilt, A. R. 1980. Archeological investigations in specified areas of the Hanalei National Wildlife Refuge, Hanalei Valley, Kaua’i, Bishop Museum, Honolulu, HI.

126

Schwartz, C. W. and E. R. Schwartz. 1952. The Hawaiian coot. The Auk 69:446-447.

Scott, D. A. (ed.). 1993. A Directory of Wetlands in Oceania. IWRB, Slimbridge, U.K. and AWB, Kuala Lumpur, Malaysia.

Shallenberger, R. J. 1977. An ornithological survey of Hawaiian wetlands. Volume I and II. Contract DACW 84-77-C-0036. Prepared for the U.S. Army Corps of Engineers, Engineer district, Honolulu, HI, 131pp and 278pp.

Stutzenbaker, C. D. and M. W. Weller. 1989. The Texas coast. Pages 385-405 in Habitat management for migrating and wintering waterfowl in North America. (Smith L. M., R. L. Pederson, and R. M. Kaminski., eds.). Texas Tech University Press, Lubbock, TX, 574pp.

Swedberg, G. E. 1967. The Koloa: a preliminary report on the life history and status of the Hawaiian duck (Anas wylvilliana). Prepared as part of the Koloa Restoration Project. Supported by: The World Wildlife Fund and the Federal Aid to Wildlife Restoration Act (W-5-R). Wildlife Branch, Division of Fish and Game, Department of Land and Natural Resources. Honolulu, HI, 57pp.

Takano, L. L. and S. M. Haig. 2004. Seasonal movement and home range of the Mariana Common Moorhen. Condor 106:652-663.

Telfer, T. C. 1973. Field investigation of native Hawaiian waterbirds on the island of Kaua’i. (S.I.). Progress report, Job. No. VIII-C(3), Project no. W-15-3 Division of Forestry and Wildlife, Honolulu, HI.

U. S. Department of Agriculture. 2000, December 6. Plant guide: Knotgrass (Paspalum distichum L.). Plant Symbol = PADI6. (2006, October 30).

U. S. Environmental Protection Agency. 2003. Wetlands: status and trends. (2003, February 24).

U. S. Fish and Wildlife Service. 1985. Hawaiian Waterbirds Recovery Plan. U.S. Fish and Wildlife Service. Portland, OR, 99pp.

U. S. Fish and Wildlife Service. 2005. Draft Revised Recovery Plan for Hawaiian Waterbirds, Second Draft of Second Revision. U.S. Fish and Wildlife Service. Portland, OR, 155pp.

Vogel, J. A. 1999. Migration chronology and habitat use of webless migratory game birds in lower Missouri river floodplain wetlands. M.S. Thesis. University of Missouri-Columbia, MO, 79pp.

127

Waller, R. A. and D. B. Duncan. 1969. A Bayes rule for the symmetric multiple comparisons problem. Journal of the American Statistical Association 64:1484- 1503.

Walther, C. R. and W. L. Hohman. 1999. Factors affecting vocalization rates of common moorhens. Page 50 in Proceedings of the marsh bird monitoring workshop, Laurel Maryland, April 1999. U.S. Fish & Wildlife Service and U.S. Geological Survey, Laurel, MD.

Wang, J. A. 1983. Introduction. Pages 1-13 in J.A. Wang (ed.). Taro, a review of Colocasia esculenta and its potentials. University of Hawaii Press, Honolulu, HI.

Weller, M. W. and L. H. Fredrickson. 1973. Avian ecology of a managed glacial marsh. Living Birds 12:269-291.

Westerkov, K. 1950. Methods for determining the age of game bird eggs. Journal of Wildlife Management 14:56-67.

Williams, F. X. 1944. Biological studies in Hawaiian water-loving insects. Part III Diptera or flies. D. Culicidae, Chironomidae and Certopogonidae. Proceedings of the Hawaiian Entomological Society 12:149-180.

Wood, N. A. 1974. Breeding behavior and biology of the Moorhen. British Birds 67:104-115.

Zar, J. H. 1984. Biostatistical analysis. Second edition, Prentice-Hall, Englewood Cliffs, NJ, 718pp.

128

APPENDIX 1

Behavior, habitat, and water depth variables recorded during visual scan surveys.

Behavior: Describes the activity of the birds at time of detection 1. Foraging 2. Loafing 3. Locomotion (swimming or walking) 4. Thermal/escape cover 5. Preening 6. Display 7. Nesting/brooding 8. Aggression 9. Other 10. Cannot determine 11. Alert

Habitat: Describes the habitat that the birds was using when detected. 1. Dry mud 2. Wet mud 3. Mud/water interface 4. Open water 5. Residual vegetation 6. Robust emergent vegetation 7. Other emergent vegetation 8. Floating vegetation (Azolla filiculoides and Lemna perpusilla) 9. Flooded shrub-scrub 10. Snag

Water depth: Water depth at site of detection 1. Dry mud 2. Wet mud 3. Mud/water interface 4. Foot 5. Up to joint 6. Joint to body 7. Body 8. Swimming 9. Cannot determine 10. None

129

APPENDIX 2

Daily precipitation recorded at Princeville Ranch, Hawai’i from 1 January to 30 November 2004 (National Climatic Data Center).

Precipitation Precipitation Precipitation Date (cm) Date (cm) Date (cm) 1/1/2004 1.52 2/1/2004 0.00 3/1/2004 4.06 1/2/2004 0.00 2/2/2004 0.00 3/2/2004 0.00 1/3/2004 0.00 2/3/2004 0.00 3/3/2004 0.00 1/4/2004 0.00 2/4/2004 0.00 3/4/2004 3.05 1/5/2004 0.00 2/5/2004 0.25 3/5/2004 0.51 1/6/2004 0.00 2/6/2004 0.00 3/6/2004 0.00 1/7/2004 0.00 2/7/2004 1.52 3/7/2004 0.76 1/8/2004 0.00 2/8/2004 0.76 3/8/2004 0.00 1/9/2004 0.00 2/9/2004 0.00 3/9/2004 0.00 1/10/2004 0.00 2/10/2004 1.02 3/10/2004 0.00 1/11/2004 0.00 2/11/2004 0.25 3/11/2004 0.25 1/12/2004 1.02 2/12/2004 0.00 3/12/2004 0.51 1/13/2004 0.00 2/13/2004 0.00 3/13/2004 5.33 1/14/2004 1.27 2/14/2004 0.00 3/14/2004 0.00 1/15/2004 0.00 2/15/2004 0.51 3/15/2004 0.00 1/16/2004 0.00 2/16/2004 0.00 3/16/2004 0.51 1/17/2004 0.00 2/17/2004 0.00 3/17/2004 0.00 1/18/2004 0.00 2/18/2004 0.00 3/18/2004 0.00 1/19/2004 0.00 2/19/2004 0.00 3/19/2004 0.00 1/20/2004 0.25 2/20/2004 0.00 3/20/2004 0.25 1/21/2004 0.76 2/21/2004 0.00 3/21/2004 0.76 1/22/2004 2.03 2/22/2004 0.00 3/22/2004 0.00 1/23/2004 0.00 2/23/2004 0.00 3/23/2004 1.02 1/24/2004 0.00 2/24/2004 0.00 3/24/2004 0.76 1/25/2004 3.56 2/25/2004 0.25 3/25/2004 0.76 1/26/2004 0.00 2/26/2004 1.27 3/26/2004 4.06 1/27/2004 0.00 2/27/2004 6.10 3/27/2004 0.00 1/28/2004 0.00 2/28/2004 0.25 3/28/2004 0.00 1/29/2004 0.00 2/29/2004 0.25 3/29/2004 1.02 1/30/2004 0.00 3/30/2004 1.27 1/31/2004 0.00 3/31/2004 0.00

130

Daily precipitation recorded at Princeville Ranch, Hawai’i from 1 January to 30 November 2004 (National Climatic Data Center)…continued.

Precipitation Precipitation Precipitation Date (cm) Date (cm) Date (cm) 4/1/2004 0.00 5/1/2004 0.76 6/1/2004 0.25 4/2/2004 1.02 5/2/2004 0.51 6/2/2004 0.25 4/3/2004 1.02 5/3/2004 0.00 6/3/2004 0.51 4/4/2004 0.51 5/4/2004 0.51 6/4/2004 0.00 4/5/2004 0.00 5/5/2004 0.25 6/5/2004 0.00 4/6/2004 0.00 5/6/2004 4.57 6/6/2004 1.52 4/7/2004 0.00 5/7/2004 0.25 6/7/2004 0.00 4/8/2004 0.00 5/8/2004 0.76 6/8/2004 1.02 4/9/2004 0.00 5/9/2004 1.52 6/9/2004 0.00 4/10/2004 0.00 5/10/2004 0.00 6/10/2004 0.00 4/11/2004 0.00 5/11/2004 1.27 6/11/2004 0.00 4/12/2004 1.02 5/12/2004 1.02 6/12/2004 0.00 4/13/2004 0.76 5/13/2004 0.00 6/13/2004 0.25 4/14/2004 0.76 5/14/2004 0.25 6/14/2004 1.52 4/15/2004 0.51 5/15/2004 2.54 6/15/2004 0.25 4/16/2004 2.03 5/16/2004 0.00 6/16/2004 0.25 4/17/2004 0.00 5/17/2004 0.00 6/17/2004 0.25 4/18/2004 0.00 5/18/2004 0.00 6/18/2004 1.78 4/19/2004 0.51 5/19/2004 0.00 6/19/2004 1.02 4/20/2004 1.27 5/20/2004 0.25 6/20/2004 0.76 4/21/2004 1.02 5/21/2004 0.00 6/21/2004 0.25 4/22/2004 0.25 5/22/2004 0.00 6/22/2004 0.51 4/23/2004 0.25 5/23/2004 0.00 6/23/2004 0.51 4/24/2004 0.00 5/24/2004 0.00 6/24/2004 0.00 4/25/2004 0.00 5/25/2004 0.00 6/25/2004 0.00 4/26/2004 0.00 5/26/2004 0.00 6/26/2004 0.00 4/27/2004 0.00 5/27/2004 0.25 6/27/2004 0.00 4/28/2004 0.51 5/28/2004 0.25 6/28/2004 0.00 4/29/2004 0.00 5/29/2004 0.00 6/29/2004 0.00 4/30/2004 3.05 5/30/2004 0.25 6/30/2004 0.00 5/31/2004 0.00

131

Daily precipitation recorded at Princeville Ranch, Hawai’i from 1 January to 30 November 2004 (National Climatic Data Center)…continued.

Precipitation Precipitation Precipitation Date (cm) Date (cm) Date (cm) 7/1/2004 0.00 8/1/2004 0.00 9/1/2004 0.00 7/2/2004 0.00 8/2/2004 0.00 9/2/2004 0.00 7/3/2004 0.00 8/3/2004 0.00 9/3/2004 0.00 7/4/2004 0.00 8/4/2004 0.25 9/4/2004 0.00 7/5/2004 0.00 8/5/2004 0.00 9/5/2004 0.00 7/6/2004 0.00 8/6/2004 0.00 9/6/2004 0.00 7/7/2004 0.00 8/7/2004 1.78 9/7/2004 0.00 7/8/2004 0.00 8/8/2004 0.00 9/8/2004 0.00 7/9/2004 0.00 8/9/2004 0.25 9/9/2004 0.00 7/10/2004 0.00 8/10/2004 0.00 9/10/2004 0.00 7/11/2004 0.00 8/11/2004 0.76 9/11/2004 0.00 7/12/2004 0.00 8/12/2004 0.51 9/12/2004 0.00 7/13/2004 0.00 8/13/2004 0.25 9/13/2004 0.00 7/14/2004 0.00 8/14/2004 0.25 9/14/2004 0.00 7/15/2004 0.00 8/15/2004 0.00 9/15/2004 0.00 7/16/2004 0.00 8/16/2004 0.00 9/16/2004 0.00 7/17/2004 0.00 8/17/2004 0.00 9/17/2004 0.00 7/18/2004 0.00 8/18/2004 0.00 9/18/2004 0.00 7/19/2004 0.00 8/19/2004 0.00 9/19/2004 0.00 7/20/2004 0.00 8/20/2004 0.00 9/20/2004 0.00 7/21/2004 0.00 8/21/2004 0.76 9/21/2004 0.00 7/22/2004 0.00 8/22/2004 0.25 9/22/2004 0.00 7/23/2004 0.00 8/23/2004 0.25 9/23/2004 0.00 7/24/2004 0.00 8/24/2004 0.51 9/24/2004 0.00 7/25/2004 0.00 8/25/2004 1.02 9/25/2004 0.00 7/26/2004 0.00 8/26/2004 0.00 9/26/2004 0.00 7/27/2004 0.00 8/27/2004 0.51 9/27/2004 0.00 7/28/2004 0.00 8/28/2004 0.25 9/28/2004 0.00 7/29/2004 0.00 8/29/2004 0.51 9/29/2004 0.00 7/30/2004 0.00 8/30/2004 0.51 9/30/2004 0.00 7/31/2004 0.00 8/31/2004 0.00

132

Daily precipitation recorded at Princeville Ranch, Hawai’i from 1 January to 30 November 2004 (National Climatic Data Center)…continued.

Precipitation Precipitation Date (cm) Date (cm) 10/1/2004 0.00 11/1/2004 0.00 10/2/2004 0.00 11/2/2004 0.00 10/3/2004 0.51 11/3/2004 0.00 10/4/2004 0.76 11/4/2004 0.00 10/5/2004 0.25 11/5/2004 0.00 10/6/2004 0.76 11/6/2004 0.00 10/7/2004 2.54 11/7/2004 0.00 10/8/2004 0.25 11/8/2004 0.00 10/9/2004 0.25 11/9/2004 0.00 10/10/2004 0.00 11/10/2004 0.00 10/11/2004 0.00 11/11/2004 0.00 10/12/2004 0.00 11/12/2004 0.00 10/13/2004 0.00 11/13/2004 0.00 10/14/2004 0.00 11/14/2004 0.00 10/15/2004 1.78 11/15/2004 0.00 10/16/2004 0.00 11/16/2004 0.00 10/17/2004 0.51 11/17/2004 0.00 10/18/2004 0.25 11/18/2004 0.00 10/19/2004 1.52 11/19/2004 0.00 10/20/2004 0.76 11/20/2004 0.00 10/21/2004 0.00 11/21/2004 0.00 10/22/2004 0.00 11/22/2004 0.00 10/23/2004 0.51 11/23/2004 0.00 10/24/2004 3.81 11/24/2004 0.00 10/25/2004 1.27 11/25/2004 0.00 10/26/2004 4.32 11/26/2004 0.00 10/27/2004 4.06 11/27/2004 0.00 10/28/2004 1.02 11/28/2004 0.00 10/29/2004 1.27 11/29/2004 0.00 10/30/2004 2.03 11/30/2004 0.00 10/31/2004 0.25

133

APPENDIX 3

Plants found at refuge wetland microhabitat sites at Hanalei NWR from 23 June to 17 November 2004 (www.plants.usda/gov/).

N = Native Wetland E = Exotic/ Indicator Scientific name Family English name naturalized Status Ammannia coccinea Lythraceae valley redstem E OBL* Bacopa monnieri Scrophulariaceae herb of grace N OBL Commelina diffusa Commelinaceae climbing dayflower E FACW** Cuphea hyssopifolia Lythraceae false heather E Cyperus polystachyos Cyperaceae manyspike flatsedge N FAC*** Echinochloa crus-galli Poaceae barnyard grass E FACW Fimbristylis dichotoma Cyperaceae forked fimbry N FAC Fimbristylis littoralis Cyperaceae fimbry FACW Ludwigia octovalvis Onagraceae Mexican primrose-willow E OBL Ludwigia palustris Onagraceae marsh seedbox E OBL Mimosa pudica Fabaceae shameplant E FACU**** Paspalum distichum Poaceae knotgrass FACW Paspalum urvillei Poaceae Vasey's grass E FACW Rhynchospora corymbosa Cyperaceae matamat OBL Schoenoplectus juncoides Cyperaceae rock bulrush N OBL Urochloa mutica Poaceae California grass, para grass E FACW *OBL (Obligate Wetland) - Occurs almost always (estimated probability 99%) under natural conditions in wetlands. **FACW (Facultative Wetland) - Usually occurs in wetlands (estimated probability 67%-99%), but occasionally found in non- wetlands. ***FAC (Facultative) - Equally likely to occur in wetlands or non-wetlands (estimated probability 34%-66%). ****FACU (Facultative Upland) - Usually occurs in non-wetlands (estimated probability 67%-99%), but occasionally found on wetlands (estimated probability 1%-33%).

134

0 (0.0%) 0 (0.0%) 5 (1.6%) 3 (1.0%) 0 (0.0%) 0 (0.0%) 1 (0.3%) 0 (0.0%) 0 (0.0%) = 311) 12 (3.9%) 12 (3.9%)

91 (29.3%) 91 (29.3%) n 199 (64.0%) 199 (64.0%) ( Hawaiian stilt

0 (0.0%) 1 (0.3%) 1 (0.3%) 3 (1.0%) 9 (3.0%) 2 (0.7%) 0 (0.0%) 2 (0.7%) 0 (0.0%) = 299) 13 (4.3%) 13 (4.3%)

duck n 160 (53.5%) 160 (53.5%) 109 (36.5%) ( Hawaiian

observations (number 1”, “B4”, and “B6”) following

1 (0.4%) 1 (0.4%) 1 (0.4%) 0 (0.0%) 2 (0.8%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 3 (1.3%) = 237) 10 (4.2%) 10 (4.2%) 78 (32.9%) 78 (32.9%) n 142 (59.9%) 142 (59.9%) (

Hawaiian coot

coot, duck, and stilt 0 (0.0%) 0 (0.0%) 1 (3.4%) 1 (3.4%)

APPENDIX 4

4 (13.8%) 4 (13.8%) 9 (31.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (6.9%) 0 (0.0%) 13 (44.8%) 13 (44.8%)

= 29) n (

refuge wetlands (“A1”, “B moorhen

Hawaiian common

2004 at Hanalei NWR.

Habitat

Habitat type of Hawaiian common moorhen, observations/ % of observations) in rototiller treatments in Dry mud Wet mud Mud/water interface Open water Residual vegetation Residual emergent vegetation Robust emergent vegetation Other emergent vegetation Floating vegetation Flooded shrub-scrub Snag

135

9 (2.9%) 9 (2.9%) 3 (1.0%) 0 (0.0%) 0 (0.0%) 1 (0.3%) 1 (0.3%) 0 (0.0%) 0 (0.0%) = 311) 10 (3.2%) 10 (3.2%) 32 (10.3%) 32 (10.3%) n 255 (82.0%) 255 (82.0%) ( Hawaiian stilt r observations/

0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (0.7%) 0 (0.0%) 9 (3.0%) 1 (0.3%) wing rototiller treatments treatments wing rototiller = 299) 29 (9.7%) 29 (9.7%) duck 61 (20.4%) 61 (20.4%) 53 (17.7%) n 144 (48.2%) 144 (48.2%) ( Hawaiian

ervations (numbe 4 (1.7%) 4 (1.7%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 3 (1.3%) 0 (0.0%) 0 (0.0%) 1 (0.4%) = 237) 20 (8.4%) 20 (8.4%) 40 (16.9%) 40 (16.9%) n 169 (71.3%) 169 (71.3%) ( Hawaiian coot

1”, “B4”, and “B6”) follo 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (3.4%)

APPENDIX 5 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 18 (62.1%) 18 (62.1%) 10 (34.5%) = 29) n ( orhen, coot, duck, and stilt obs moorhen Hawaiian common Behavior Behavior of Hawaiian common mo % of observations) in refuge wetlands (“A1”, “B NWR. in 2004 at Hanalei Foraging Loafing Locomotion Thermal/ escape cover Preening Display Nesting/ brooding Aggression Other Cannot determine Alert

136

11 (3.5%) 9 (2.9%) 9 (2.9%) 1 (0.3%) 8 (2.6%) 8 (2.6%) 5 (1.6%) = 311) 10 (3.2%) 10 (3.2%) 21 (6.8%) 13 (4.2%) 13 (4.2%) 11 (3.5%) n 222 (71.4%) 222 (71.4%) ( Hawaiian stilt

0 (0.0%) 0 (0.0%) 3 (1.0%) 3 (1.0%) 4 (1.3%) = 299) 34 (11.4%) 34 (11.4%) 13 (4.3%) 13 (4.3%) 24 (8.0%) 14 (4.7%) 17 (5.7%) duck 59 (19.7%) 59 (19.7%) n 131 (43.8%) 131 (43.8%) ( Hawaiian

3 (1.3%) 3 (1.3%) 4 (1.7%) 4 (1.7%) 6 (2.5%) 4 (1.7%) 5 (2.1%) 6 (2.5%) = 237) 13 (5.5%) 13 (5.5%) 22 (9.3%) 49 (20.7%) 49 (20.7%) n 125 (52.7%) 125 (52.7%) ( Hawaiian coot

common moorhen, coot, duck, and stilt observations 0 (0.0%) 0 (0.0%) 1 (3.4%) 1 (3.4%)

APPENDIX 6 2 (6.9%) 2 (6.9%) 5 (17.2%) 5 (17.2%) 7 (24.1%) 2 (6.9%) 2 (6.9%) 0 (0.0%) 4 (13.8%) 4 (13.8%) 7 (24.1%) = 29) n ( tions) in refuge wetlands (“A1”, “B1”, “B4”, and “B6”) following moorhen Hawaiian common 2004 at Hanalei NWR. Water depth Dry mud mud Wet interface Mud/water Foot Up to joint body Joint to Body Swimming Cannot determine None Water depth at site of detection for Hawaiian (number observations/ % of observa rototiller treatments in

137

APPENDIX 7

Area of taro (ha) in 8 habitat categories at Hanalei NWR from 3 March to 12 November 2004. Habitat categories in taro lo’i

Scan survey date Recently planted growth Early mature or to Medium dense growth Lo’i being harvested Unvegetated wet fallow wet Vegetated fallow Unvegetated dry fallow dry Vegetated fallow Mar. 3/5 5.844 9.366 19.800 2.906 6.346 1.594 2.079 1.406 Mar. 8/9 5.626 8.576 20.623 2.788 5.190 1.286 3.171 2.074 Mar. 15/17 5.907 7.636 20.781 3.676 4.788 0.964 3.886 1.703 Mar. 26/27 6.676 6.772 21.276 3.458 4.618 2.577 2.941 1.024 April 5 6.922 7.086 20.512 3.289 5.351 2.023 2.149 2.009 April 10 7.430 7.746 20.290 3.041 4.050 1.167 2.301 3.316 April 17/19 7.676 7.767 19.854 3.204 3.480 1.846 3.195 2.320 April 24 7.139 9.125 19.013 2.936 4.189 1.920 2.925 2.094 May 1/2 6.636 9.194 19.125 2.985 4.441 2.227 3.300 1.435 May 8 6.371 9.397 18.995 2.840 4.170 3.001 3.276 1.290 May 14 6.651 8.731 18.801 3.354 3.766 2.088 4.002 1.947 May 21 6.848 9.668 18.197 3.476 3.576 2.039 3.185 2.351 May 26 6.583 9.961 17.963 3.829 2.914 1.587 3.961 2.544 June 1 6.414 10.044 17.782 3.590 2.677 1.216 4.769 2.611 June 16/17 6.431 9.862 18.155 4.234 3.282 0.490 3.589 3.059 June 26 7.523 9.398 17.117 4.800 3.082 1.537 3.203 2.441 July 2 7.364 9.261 17.341 3.852 2.475 1.980 4.097 2.730 July 8 7.062 9.942 16.730 2.892 2.917 2.137 5.126 2.295 July 15 6.466 10.018 16.864 2.661 2.739 1.859 5.126 3.368 July 23 6.267 10.528 15.269 3.242 2.702 2.193 6.518 2.381 July 30 6.907 10.339 14.714 3.265 2.828 2.243 6.855 1.950 Aug. 5 6.969 10.240 14.456 3.519 4.400 1.591 4.758 3.167 Sept. 16 8.175 10.882 13.773 4.114 3.165 0.751 4.173 4.307 Sept. 23 8.465 10.998 13.773 3.952 3.646 0.973 3.013 4.520 Sept. 30 6.540 12.702 13.378 3.292 4.064 0.579 4.998 3.787 Oct. 8 6.717 13.083 13.519 3.106 3.593 0.475 4.882 3.965 Oct. 15 6.717 13.083 13.519 3.106 3.593 0.475 4.882 3.965 Nov. 12 5.704 15.461 10.655 4.072 5.626 1.042 2.667 4.113

138

APPENDIX 8

Endangered waterbird observations (total number of observations/%) by taro habitat category at Hanalei NWR from 3 March to 12 November 2004. Hawaiian Hawaiian Hawaiian Hawaiian moorhen coot duck stilt (n = 2454) (n = 775) (n = 7776) (n = 3917) Recently planted 139 (5.7%) 34 (4.4%) 378 (4.9%) 1009 (25.8%) Early or mature growth 64 (2.6%) 47 (6.1%) 52 (0.7%) 117 (3.0%) Medium to dense growth 24 (1.0%) 30 (3.9%) 61 (0.8%) 74 (1.9%) Lo’i being harvested 201 (8.2%) 131 (16.9%) 540 (6.9%) 455 (11.6%) Unvegetated wet fallow 194 (7.9%) 157 (20.3%) 835 (10.7%) 1407 (35.9%) Vegetated wet fallow 84 (3.4%) 40 (5.2%) 265 (3.4%) 215 (5.5%) Unvegetated dry fallow 45 (1.8%) 6 (0.8%) 68 (0.9%) 173 (4.4%) Vegetated dry fallow 36 (1.5%) 36 (4.6%) 104 (1.3%) 39 (1.0%) Taro dikes 1667 (67.9%) 294 (37.9%) 5473 (70.4%) 428 (10.9%)

139

0 (0.0%) 2 (5.6%) 1 (2.8%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 4 (11.1%) = 36) 29 (80.6%) n fallow ( Vegetated dry Vegetated

2 (4.4%) 2 (4.4%) 4 (8.9%) 0 (0.0%) 2 (4.4%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (2.2%) 0 (0.0%) 0 (0.0%) = 45) 36 (80.0%) 36 (80.0%) n fallow ( Unvegetated dry Unvegetated

5 (6.0%) 5 (6.0%) 0 (0.0%) 0 (0.0%) 2 (2.4%) 1 (1.2%) 0 (0.0%) 0 (0.0%) 2 (2.4%) = 84) 51 (60.7%) 51 (60.7%) 11 (13.1%) 12 (14.3%) n fallow ( Vegetated wet Vegetated

tions) in taro dikes and habitat

3 (1.5%) 3 (1.5%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (0.5%) 2 (1.0%) 15 (7.7%) 15 (7.7%) 14 (7.2%) = 194) 26 (13.4%) 26 (13.4%) fallow 133 (68.6%) 133 (68.6%) n ( Unvegetated wet

2 (1.0%) 2 (1.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (0.5%) 2 (1.0%) 15 (7.5%) 15 (7.5%) 13 (6.5%) = 201) 31 (15.4%) 31 (15.4%) 137 (68.2%) 137 (68.2%) n ( harvested Lo’i being Habitat categories in taro lo’i in taro Habitat categories

0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (4.2%) 0 (0.0%) 1 (4.2%) 0 (0.0%) 0 (0.0%) 1 (4.2%) 3 (12.5%) 3 (12.5%) 4 (16.7%) = 24) 14 (58.3%) 14 (58.3%) n ( APPENDIX 9 Medium to dense growth dense growth

1 (1.6%) 1 (1.6%) 0 (0.0%) 0 (0.0%) 5 (7.8%) 2 (3.1%) 1 (1.6%) 0 (0.0%) 1 (1.6%) 9 (14.1%) 9 (14.1%) 2 (31.0%) = 64) ved (number observations/ % of observa 43 (67.2%) 43 (67.2%) n mature mature ( Early growth or or Early growth

March to 12 November 2004. 1 (0.7%) 1 (0.7%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 3 (2.2%) 0 (0.0%) 0 (0.0%) 0 (0.0%) = 139) = 139) 18 (12.9%) 18 (12.9%) 29 (20.9%) 88 (63.3%) 88 (63.3%) n planted Recently (

2 (0.1%) 2 (0.1%) 0 (0.0%) 7 (0.4%) 0 (0.0%) 34 (2.0%) 34 (2.0%) 70 (4.2%) 67 (4.0%) = 1667) 503 (30.2%) 503 (30.2%) 173 (10.4%) 414 (24.8%) 397 (23.8%) n ( Taro dikes ories at Hanalei NWR from 3 ories at Hanalei NWR g Behavior Foraging Loafing Locomotion escape Thermal/ cover Preening Display Nesting/ brooding Aggression Other Cannot determine Alert Hawaiian common moorhen behavior obser cate

140

1 (2.8%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (5.6%) 0 (0.0%) 0 (0.0%) 0 (0.0%) = 36) 4 (11.1%) n fallow fallow 29 (80.6%) ( Vegetated dry Vegetated

= 6) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 3 (50.0%) 1 (16.7%) 2 (33.3%) n ( dry fallow dry fallow Unvegetated bitat categories at Hanalei

0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (2.5%) 0 (0.0%) 0 (0.0%) 0 (0.0%) = 40) 5 (12.5%) 7 (17.5%) n fallow 27 (67.5%) ( Vegetated wet wet Vegetated

2 (1.3%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) = 157) 17 (10.8%) 26 (16.6%) n 112 (71.3%) ( wet fallow wet fallow Unvegetated

1 (0.8%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (0.8%) vations) in taro dikes and ha = 131) 13 (9.9%) 94 (71.8%) 22 (16.8%) n ( harvested harvested Lo’i being

Habitat categories in taro lo’i in taro Habitat categories

1 (3.3%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) = 30) 4 (13.3%) 6 (20.0%) n 19 (63.3%) ( APPENDIX 10 Medium to dense growth dense growth

0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (2.1%) = 47) n 21 (44.7%) 12 (25.5%) 13 (27.7%) ( or mature mature or Early growth Early growth

(number observations/ % of obser 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) = 34) = 34) 9 (26.5%) 6 (17.6%) n 19 (55.9%) planted ( Recently Recently

0 (0.0%) 0 (0.0%) 0 (0.0%) 5 (1.7%) 1 (0.3%) 0 (0.0%) 6 (2.0%) = 294) 94 (32.0%) 30 (10.2%) 48 (16.3%) n 110 (37.4%) ( Taro dikes Taro dikes Behavior WR from 3 March to 12 November 2004. Foraging Loafing Locomotion escape Thermal/ cover Preening Display Nesting/ brooding Aggression Other Cannot determine Alert N Hawaiian coot behavior observed

141

0 (0.0%) 8 (7.7%) 0 (0.0%) 0 (0.0%) 1 (1.0%) 0 (0.0%) 1 (1.0%) 1 (1.0%) = 104) fallow fallow 51 (49.0%) 17 (16.3%) 25 (24.0%) n ( Vegetated dry Vegetated

3 (4.4%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (2.9%) = 68) n 12 (17.6%) 37 (54.4%) 14 (20.6%) ( dry fallow dry fallow Unvegetated

0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 4 (1.5%) = 265) fallow fallow 34 (12.8%) 29 (10.9%) 32 (12.1%) n 166 (62.6%) ( Vegetated wet wet Vegetated

ro dikes and habitat categories at 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (0.2%) 0 (0.0%) 4 (0.5%) = 835) 44 (5.3%) 12 (1.4%) n 537 (64.3%) 105 (12.6%) 131 (15.7%) ( wet fallow wet fallow Unvegetated

2 (0.4%) 0 (0.0%) 1 (0.2%) 0 (0.0%) 0 (0.0%) 2 (0.4%) 2 (0.4%) = 540) 35 (6.5%) 59 (10.9%) 91 (16.9%) n 348 (64.4%) ( harvested harvested Lo’i being

Habitat categories in taro lo’i in taro Habitat categories

4 (6.6%) 3 (4.9%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) = 61) n 27 (44.3%) 10 (16.4%) 17 (27.9%) ( APPENDIX 11 Medium to dense growth dense growth

2 (3.8%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 5 (9.6%) 0 (0.0%) = 52) n 19 (36.5%) 11 (21.2%) 15 (28.2%) r observations/ % of observations) in ta ( or mature mature or Early growth Early growth

0 (0.0%) 0 (0.0%) 4 (1.1%) 0 (0.0%) 1 (0.3%) 0 (0.0%) 1 (0.3%) = 378) = 378) 17 (4.5%) 43 (11.4%) 44 (11.6%) n planted 268 (70.9%) Recently Recently (

2 (0.0%) 0 (0.0%) 7 (0.1%) 1 (0.0%) 72 (1.3%) 20 (0.4%) 15 (0.3%) = 5473) 372 (6.8%) 113 (2.1%) 843 (15.4%) n ( 4028 (73.6%) Taro dikes Taro dikes Behavior Foraging Loafing Locomotion escape Thermal/ cover Preening Display Nesting/ brooding Aggression Other Cannot determine Alert Hanalei NWR from 3 March to 12 November 2004. Hanalei NWR from 3 March to 12 November Hawaiian duck behavior observed (numbe

142

0 (0.0%) 0 (0.0%) 3 (7.7%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) = 39) n fallow fallow 24 (61.5%) 12 (30.8%) ( Vegetated dry Vegetated

9 (5.2%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 3 (1.7%) 0 (0.0%) 0 (0.0%) 1 (0.6%) = 173) 66 (38.2%) 70 (40.5%) 24 (13.9%) n ( dry fallow dry fallow Unvegetated

2 (0.9%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (0.9%) = 215) 20 (9.3%) fallow fallow 32 (14.9%) n 159 (74.0%) ( Vegetated wet wet Vegetated

dikes and habitat categories at Hanalei 1 (0.1%) 0 (0.0%) 7 (0.5%) 7 (0.5%) 0 (0.0%) 0 (0.0%) 8 (0.6%) 13 (0.9%) = 1407) n 976 (69.4%) 182 (12.9%) 213 (15.1%) ( wet fallow wet fallow Unvegetated

3 (0.7%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (0.4%) 4 (0.9%) = 455) 11 (2.4%) 41 (9.0%) 55 (12.1%) n 339 (74.5%) ( harvested harvested Lo’i being

Habitat categories in taro lo’i in taro Habitat categories

5 (6.8%) 0 (0.0%) 0 (0.0%) 2 (2.7%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) = 74) 8 (10.8%) n 44 (59.5%) 15 (20.3%) ( APPENDIX 12 Medium to dense growth dense growth

7 (6.0%) 4 (3.4%) 2 (1.7%) 7 (6.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (0.9%) = 117) observations/ % of observations) in taro 96 (82.1%) n ( or mature mature or Early growth Early growth

1 (0.1%) 0 (0.0%) 0 (0.0%) 3 (0.3%) 2 (0.2%) 0 (0.0%) 4 (0.4%) 97 (9.6%) 15 (1.5%) = 1009) = 1009) planted n 749 (74.2%) 138 (13.7%) Recently Recently (

1 (0.2%) 0 (0.0%) 3 (0.7%) 9 (2.1%) 3 (0.7%) 2 (0.5%) 34 (7.9%) = 428) 67 (15.7%) 60 (14.0%) 90 (21.0%) n 159 (37.1%) ( Taro dikes Taro dikes Behavior WR from 3 March to 12 November 2004. Foraging Loafing Locomotion escape Thermal/ cover Preening Display Nesting/ brooding Aggression Other Cannot determine Alert N Hawaiian stilt behavior observed (number

143

1 (2.8%) 1 (2.8%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) = 36) 6 (16.7%) 6 (16.7%) 4 (11.1%) n fallow fallow 16 (44.4%) ( Vegetated dry Vegetated

2 (4.4%) 3 (6.7%) 1 (2.2%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) = 45) 7 (15.6%) n 18 (40.0%) 14 (31.1%) ( dry fallow dry fallow Unvegetated

number observations/ % of 0 (0.0%) 5 (6.0%) 0 (0.0%) 4 (4.8%) 6 (7.1%) 3 (3.6%) 4 (4.8%) = 84) n fallow fallow 20 (23.8%) 11 (13.1%) 31 (36.9%) ( Vegetated wet wet Vegetated

0 (0.0%) 9 (4.6%) 6 (3.1%) 3 (1.5%) 4 (2.1%) = 194) 13 (6.7%) 17 (8.8%) 15 (7.7%) 63 (32.5%) 64 (33.0%) n ( wet fallow wet fallow Unvegetated

m 3 March to 12 November 2004.

0 (0.0%)

= 201) 17 (8.5%) 11 (5.5%) 0 (0.0%) 2 (1.0%) 0 (0.0%) 7 (3.5%) 62 (30.8%) 12 (6.0%) n ( 47 (23.4%) harvested harvested Lo’i being

Habitat categories in taro lo’i in taro Habitat categories mmon moorhen observations (

APPENDIX 13 0 (0.0%) 0 (0.0%) = 24) 9 (37.5%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (4.2%) 1 (4.2%) n 9 (37.5%) 3 (12.5%) ( Medium to dense growth dense growth

1 (1.6%) 3 (4.7%) = 64) 4 (6.3%) 1 (1.6%) 0 (0.0%) 3 (4.7%) n 17 (26.6%) 31 (36.9%) 8 (12.5%) categories at Hanalei NWR fro ( 16 (25.0%) or mature mature or Early growth Early growth

5 (3.6%) 3 (1.5%) 2 (1.4%) = 139) = 139) 0 (0.0%) 0 (0.0%) 0 (0.0%) 7 (5.0%) 14 (10.1%) 28 (20.1%) n planted Recently Recently ( 76 (54.7%) Water depth Dry mud mud Wet Mud/water interface Foot Up to joint body Joint to Body Swimming Cannot determine None Water depth at site of detection for Hawaiian co observations) in taro habitat

144

= 36) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) n fallow fallow 24 (66.7%) 8 (22.2%) 4 (11.1%) ( Vegetated dry Vegetated

= 6) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) n ( 2 (33.3%) 3 (50.0%) 1 (16.7%) dry fallow dry fallow Unvegetated

1 (2.5%) = 40) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 3 (7.5%) 1 (2.5%) n fallow fallow 5 (12.5%) 4 (10.0%) ( 26 (65.0%) Vegetated wet wet Vegetated observations/ % of observations) in

1 (0.6%) 2 (1.3%) 6 (3.8%) 3 (1.9%) 0 (0.0%) 3 (1.9%) = 157) 10 (6.4%) 11 (7.0%) 19 (12.1%) n 102 (65.0%) ( wet fallow wet fallow Unvegetated

1 (0.8%) 2 (1.5%)

= 131) 1 (0.8%) 2 (1.5%) 0 (0.0%) 3 (2.3%) 5 (3.8%) 2 (1.5%) 24 (66.7%) n 115 (87.8%) ( harvested harvested Lo’i being

Habitat categories in taro lo’i in taro Habitat categories

APPENDIX 14 0 (0.0%) = 30) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (6.7%) 2 (6.7%) 0 (0.0%) n 9 (30.0%) ( 17 (56.7%) Medium to dense growth dense growth

r Hawaiian coot observations (number 1 (2.5%) = 47) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (4.3%) 1 (2.1%) 1 (2.1%) n 5 (10.6%) ( nalei NWR from 3 March to 12 November 2004 at Hanalei NWR. nalei NWR 38 (80.9%) or mature mature or Early growth Early growth

0 (0.0%) 1 (2.9%) 3 (1.9%) = 34) = 34) 7 (20.6%) 0 (0.0%) 0 (0.0%) 1 (2.9%) 0 (0.0%) n 20 (58.8%) 5 (14.7%) planted ( Recently Recently Water depth Dry mud mud Wet Mud/water interface Foot Up to joint body Joint to Body Swimming Cannot determine None taro habitat categories at Ha Water depth at site of detection fo

145

6 (5.8%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (1.9%) 0 (0.0%) 3 (2.9%) = 104) fallow fallow 26 (25.0%) 25 (24.0%) 42 (40.4%) n ( Vegetated dry Vegetated

observations) i 1 (1.5%) 4 (5.9%) 0 (0.0%) 0 (0.0%) 3 (4.4%) 3 (4.4%) 0 (0.0%) = 68) n 10 (14.7%) 26 (38.2%) 21 (30.9%) ( dry fallow dry fallow Unvegetated

= 265) 22 (8.3%) 2 (0.8%) 3 (1.1%) 4 (1.5%) 8 (3.0%) fallow fallow 16 (6.0%) 10 (3.8%) 11 (4.2%) n ( 62 (23.4%) 127 (47.9%) Vegetated wet wet Vegetated

r observations/ % of

1 (0.1%) = 835) 4 (0.5%) 0 (0.0%) 24 (2.9%) 11 (1.3%) 24 (2.9%) 21 (2.5%) 27 (3.2%) 63 (7.5%) n ( wet fallow wet fallow 660 (79.0%) Unvegetated

= 540) 8 (1.5%) 7 (1.3%) 2 (0.4%) 6 (1.1%) 42 (40.4%) 20 (3.7%) 14 (2.6%) 13 (2.4%) 23 (4.3%) n ( harvested harvested Lo’i being 437 (80.9%)

Habitat categories in taro lo’i lo’i taro in categories Habitat

APPENDIX 15 = 61) 2 (3.3%) 1 (1.6%) 3 (4.9%) 3 (4.9%) 4 (6.6%) 2 (3.3%) 1 (1.6%) n 10 (14.7%) ( 20 (32.8%) 25 (41.0%) Medium to dense growth dense growth

NWR from 3 March to 12 November 2004. NWR for Hawaiian duck observations (numbe = 52) 22 (8.3%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 3 (5.8%) 2 (3.8%) 1 (1.9%) 5 (9.6%) n ( 39 (75.0%) or mature mature or Early growth Early growth

1 (0.1%) = 378) = 378) 4 (1.1%) 0 (0.0%) 4 (1.1%) 2 (0.5%) 11 (2.9%) 17 (4.5%) 10 (2.6%) 32 (8.5%) n planted Recently Recently ( 298 (78.8%) Water depth Dry mud mud Wet Mud/water interface Foot Up to joint body Joint to Body Swimming Cannot determine None taro habitat categories at Hanalei Water depth at site of detection

146

1 (2.6%) 0 (0.0%) 0 (0.0%) 0 (0.0%) 2 (5.1%) = 39) 5 (12.8%) 5 (12.8%) 4 (10.3%) 5 (12.8%) n fallow fallow 17 (43.6%) ( Vegetated dry Vegetated

observations) in 0 (0.0%) 0 (0.0%) 0 (0.0%) 1 (0.6%) = 173) 17 (9.8%) 46 (26.6%) 42 (24.3%) 19 (11.0%) 28 (16.2%) 20 (11.6%) n ( dry fallow dry fallow Unvegetated

5 (2.3%) 6 (2.8%) 7 (3.3%) 0 (0.0%) 2 (0.9%) 4 (1.9%) 0 (0.0%) = 215) 16 (7.4%) 16 (7.4%)

fallow fallow n 159 (74.0%) ( Vegetated wet wet Vegetated observations/ % of

5 (0.4%) 1 (0.1%) 1 (0.1%) 20 (1.4%) 63 (4.5%) 49 (3.5%) 14 (1.0%) 23 (1.6%) = 1407) n 240 (17.1%) 991 (70.4%) ( wet fallow wet fallow Unvegetated ories in taro lo’i ories in taro

g 5 (1.1%) 8 (1.8%) 2 (5.1%)

= 455) 16 (3.5%) 5 (1.1%) 9 (2.0%) 0 (0.0%) 29 (6.4%) n ( 96 (21.1%) harvested harvested Lo’i being 286 (62.9%)

APPENDIX 16 Habitat cate 0 (0.0%) 0 (0.0%) = 74) 1 (1.4%) 5 (6.8%) 2 (0.0%) 5 (6.8%) 0 (0.0%) n 20 (11.6%) 9 (12.2%) ( 48 (64.9%) Medium to dense growth dense growth

r Hawaiian stilt observations (number NWR from 3 March to 12 November 2004. NWR 1 (1.6%) 0 (0.0%) = 117) 4 (6.3%) 1 (1.6%) 0 (0.0%) 3 (4.7%) 17 (26.6%) 11 (17.2%) 8 (12.5%) n ( 16 (25.0%) or mature mature or Early growth Early growth

6 (0.6%) 8 (0.8%) 3 (0.3%) 1 (0.1%) 15 (1.5%) 9 (0.9%) 6 (0.6%) 0 (0.0%) 6 (6.5%) = 1009) = 1009) planted n Recently Recently ( 895 (88.7%) Water depth Dry mud mud Wet Mud/water interface Foot Up to joint body Joint to Body Swimming Cannot determine None Water depth at site of detection fo taro habitat categories at Hanalei

147

APPENDIX 17

Plants found at EWB high density sites in taro lo’i at Hanalei NWR from 3 March to 12 November 2004 (www.plants.usda/gov/).

N = Native Wetland E = Exotic/ Indicator Scientific name Family English name naturalized Status Alternanthera sessilis Amaranthaceae sessile joyweed FAC* Ammannia coccinea Lythraceae valley redstem E OBL** Arundo donax Poaceae giant reed E FACU*** Azolla filiculoides Azollaceae Pacific mosquito fern E OBL Bacopa monnieri Scrophulariaceae herb of grace N OBL Chara spp. Characeae OBL Ceratopteris thalictroides Parkeriaceae water sprite OBL Colocasia esculenta Araceae Taro E OBL Commelina diffusa Commelinaceae climbing dayflower E FACW**** Cuphea hyssopifolia Lythraceae false heather E Cyperus difformis Cyperaceae variable flatsedge E OBL Cyperus odoratus Cyperaceae fragrant flatsedge FACW Cyperus polystachyos Cyperaceae manyspike flatsedge N FAC Echinochloa crus-galli Poaceae barnyard grass E FACW Eclipta prostrata Asteraceae false daisy E FACW Eleocharis obtusa Cyperaceae blunt spike rush N OBL Fimbristylis dichotoma Cyperaceae forked fimbry N FAC Fimbristylis littoralis Cyperaceae Fimbry FACW Kyllinga brevifolia Cyperaceae shortleaf spikesedge E FAC Lemna perpusilla Lemnaceae minute duckweed NO***** Ludwigia octovalvis Onagraceae Mexican primrose-willow E OBL Ludwigia palustris Onagraceae marsh seedbox E OBL Mimosa pudica Fabaceae shameplant E FACU Monochoria vaginalis Pontederiaceae heartshape false pickerelweed E OBL Paspalum distichum Poaceae knotgrass FACW Paspalum urvillei Poaceae Vasey's grass E FACW Rhynchospora corymbosa Cyperaceae matamat OBL Sagittaria latifolia Alismataceae broadleaf arrowhead E OBL Schoenoplectus juncoides Cyperaceae rock bulrush N OBL Urochloa mutica Poaceae California grass, para grass E FACW Wolffia globosa Lemnaceae Asian watermeal N NO *FAC (Facultative) - Equally likely to occur in wetlands or non-wetlands (estimated probability 34%-66%). **OBL (Obligate Wetland) - Occurs almost always (estimated probability 99%) under natural conditions in wetlands. ***FACU (Facultative Upland) - Usually occurs in non-wetlands (estimated probability 67%-99%), but occasionally found on wetlands (estimated probability 1%-33%). ****FACW (Facultative Wetland) - Usually occurs in wetlands (estimated probability 67%-99%), but occasionally found in non- wetlands. *****NO (No occurrence) - The species does not occur in that region.

148

2.4 0.3 2.8 ± ± ±

= 9) = 10) = 25) n None fallow fallow n n ( ( ( 6.4 3.1 5.6 Vegetated dry Vegetated

0.0 0.0 0.0 0.0

± ± ± ± dry fallow 0.0 (N/A) 0.0 (N/A) 0.0 (N/A) 0.0 (N/A) Unvegetated

2.0 1.0 2.4 0.6

± ± ± ± = 3) = 14) = 23) = 47) n fallow n n n ( ( ( ( 4.5 1.8 3.6 3.7 Vegetated wet wet Vegetated stilts at Hanalei NWR from 3 March from 3 March stilts at Hanalei NWR

0.0 0.0

0.0 0.0 ± ± ± ± (N/A) (N/A) 0.0 0.0 wet fallow wet fallow 0.0 (N/A) 0.0 (N/A) Unvegetated

1.4 2.1 1.4 184 ± ± ± ± gory measured in vegetated sample plots of taro lo’i = 61) = 26) = 46) = 27) n n n n ( ( ( ( 1.4 3.2 1.3 harvested harvested Lo’i being 2.4

Habitat categories in taro lo’i lo’i in taro Habitat categories 0.3 0.5 0.8

± APPENDIX 18 ± ± = 85) = 93) = 55) None n n n ( ( ( 0.1 0.3 Medium to 0.05 dense growth dense growth

0.4 2.0 1.4 1.4 n common moorhens, coots, ducks, and ± ± ± ± = 7) = 99) = 58) = 129) n n n ( n ( SD; n) by taro habitat cate SD; n) by taro habitat ( 1.4 1.3 0.2 0.9 or mature mature or ( ± Early growth Early growth

1.5 0.0 0.0 1.3 ± ± ± ± = 45) = 22) = 34) = 61) n n n n planted ( ( ( ( Recently Recently 0.8 0.0 0.0 0.5

umber of plant species (mean species umber of plant Hawaiian stilt Hawaiian coot Hawaiian duck Hawaiian common moorhen N to 12 November 2004. with the highest densities of Hawaiia densities with the highest

149

1.9 0.5 1.2 ± ± ±

= 40) = 36) = 104) None fallow fallow n n n ( ( 2.0 0.4 1.0 ( Vegetated dry Vegetated

0.0 0.0 0.0 0.0

± ± ± ± dry fallow 0.0 (N/A) 0.0 (N/A) 0.0 (N/A) 0.0 (N/A) Unvegetated

1.0 1.5 1.0 0.5

± ± ± ± = 56) = 160) = 216) = 100) fallow n n n n ( 0.8 0.6 0.6 0.2 ( ( ( Vegetated wet wet Vegetated

stilts at Hanalei NWR from 3 March from 3 March stilts at Hanalei NWR 0.0

0.0 0.0 0.0 ± ± ± ± (N/A) 0.0 wet fallow wet fallow 0.0 (N/A) 0.0 (N/A) 0.0 (N/A) Unvegetated

1.2 2.7 2.4 2.9 ± ± ± ± = 284) = 196) = 128) gory measured in vegetated sample plots of taro lo’i = 152) n n n 1.9 2.8 2.9 n harvested harvested ( ( ( Lo’i being 2.5 (

Habitat categories in taro lo’i lo’i in taro Habitat categories 1.2 2.1

2.8 ± ± APPENDIX 19 ± = 338) = 220) None = 400) n n 8.7 8.1 n ( ( Medium to 9.2 ( dense growth dense growth

3.2 3.4 2.6 2.5 ± ± ± n common moorhens, coots, ducks, and ± = 208) = 412) = 232) = 476) n n 7.4 6.8 6.3 n n or mature mature or ( ( ( ( 6.4 Early growth Early growth SD; n) by taro habitat cate SD; n) by taro habitat ±

1.2 1.6 1.5 1.9 ± ± ± ± = 180) = 180) = 300) = 92) planted n n n Recently Recently 0.7 0.8 1.0 n ( ( ( 2.1 (

Hawaiian common moorhen Hawaiian coot Hawaiian duck Hawaiian stilt Pooled visual obscurity (mean of Hawaiia densities with the highest to 12 November 2004.

150

± ± ±

= 25) = 25) = 25) 16.4 11.7 28.0 fallow fallow n n n 5.1% ( ( ( 18.2% 37.7% Vegetated dry Vegetated

± ± ±

0.0 0.0 0.0 0.0% 0.0% 0.0% dry fallow Unvegetated

± ±

= 46) = 46) = 46) y with the highest density of 9.1 24.7 24.8 fallow n n n 4.6% ( ( ( 28.6% 36.7% Vegetated wet wet Vegetated

± ± ±

0.0 0.0 0.0 0.0% 0.0% 0.0% wet fallow wet fallow Unvegetated

± ± ±

= 46) = 46) = 46) 16.9 11.9 25.3 n n n 9.3% ( ( ( 14.5% 33.8% harvested harvested Lo’i being

± ± ± Habitat categories in taro lo’i lo’i in taro Habitat categories mple plots by taro habitat categor

= 55) = 55) = 55) 6.0 4.6 28.2 n n n 8.5% 5.9% ( ( ( 44.1% APPENDIX 20 Medium to dense growth dense growth

± ± ± = 58) = 58) = 58) 8.3 2.9 15.1 n n n ( ( ( 10.3% 12.4% 58.0% or mature mature or Early growth Early growth

± ± ± from 3 March to 12 November 2004. = 34) = 34) = 34) 3.4 1.3 12.0 n n n 2.1% 1.3% planted ( ( ( Recently Recently 11.4% SD; range; n) of vegetated sa ±

(at 15 cm) Green cover (canopy) (surface) Green cover Green cover Hawaiian ducks at Hanalei NWR Percent green cover (mean

151

None None None fallow fallow Vegetated dry Vegetated

± ± ±

0.0 0.0 0.0 0.0% 0.0% 0.0% dry fallow Unvegetated

± ±

= 23) = 23) = 23) y with the highest density of 7.8 3.5 11.3 fallow n n n 5.4% 2.3% ( ( ( 10.6% Vegetated wet wet Vegetated

± ± ±

0.0 0.0 0.0 0.0% 0.0% 0.0% wet fallow wet fallow Unvegetated

± ± ±

= 27) = 27) = 27) 38.6 20.1 99.9 n n n 8.5% ( ( ( 20.1% 46.8% harvested harvested Lo’i being

Habitat categories in taro lo’i lo’i in taro Habitat categories mple plots by taro habitat categor

None None None APPENDIX 21 Medium to dense growth dense growth

± ± ± = 7) = 7) = 7) 4.6 1.0 14.9 n n n ( ( ( 5.4% 3.3% 44.0% or mature mature or Early growth Early growth

± ± ± = 61) = 61) = 61) 9.1 2.2 6.3 n n n 9.7% 1.9% 2.6% planted ( ( ( Recently Recently SD; range; n) of vegetated sa ±

(at 15 cm) Green cover (canopy) (surface) Green cover Green cover Hawaiian stilts at Hanalei NWR from 3 March to 12 November 2004. Percent green cover (mean

152

± ± ±

= 10) = 10) = 10) 36.8 26.5 32.8 fallow fallow n n n ( ( ( 19.5% 21.1% 39.1% Vegetated dry Vegetated

± ± ±

0.0 0.0 0.0 0.0% 0.0% 0.0% dry fallow Unvegetated e highest density of

± ± ±

= 14) = 14) = 14) 14.7 17.1 21.9 fallow n n n 3.9% 4.6% 5.9% ( ( ( Vegetated wet wet Vegetated

± ± ±

0.0 0.0 0.0 0.0% 0.0% 0.0% wet fallow wet fallow Unvegetated habitat category with th category habitat

± ± ±

= 65) = 65) = 65) 8.6 15.5 24.8 n n n 4.5% ( ( ( 10.4% 21.9% harvested harvested Lo’i being ch to 12 November 2004.

± ± ± Habitat categories in taro lo’i lo’i in taro Habitat categories

= 88) = 88) = 88) 7.7 APPENDIX 22 13.1 18.6 n n n 9.4% 8.8% ( ( ( 64.6% Medium to dense growth dense growth

± ± ± = 99) = 99) = 99) 4.2 4.1 15.5 n n n 5.2% 4.6% ( ( ( 34.8% or mature mature or Early growth Early growth

± ± ± SD; n) of vegetated sample plots by taro = 45) = 45) = 45) 1.2 7.0 ± 10.1 n n n 1.1% 3.1% planted ( ( ( Recently Recently 10.2%

(surface) (at 15 cm) Green cover (canopy) Percent green cover (mean 3 Ma NWR from at Hanalei moorhens Hawaiian common Green cover Green cover

153

± ± ± = 9) = 9) = 9)

4.6 16.6 27.6 n n n fallow fallow ( ( ( 3.1% 18.7% 34.3% Vegetated dry Vegetated

± ± ±

0.0 0.0 0.0 0.0% 0.0% 0.0% dry fallow Unvegetated

highest density of ± ±

= 22) = 22) = 22) 2.4 8.4 13.7 fallow n n n 3.0% 4.3% ( ( ( 11.3% Vegetated wet wet Vegetated

± ± ±

0.0 0.0 0.0 0.0% 0.0% 0.0% wet fallow wet fallow Unvegetated itat category with the category itat

± ± ±

= 25) = 25) = 25) 5.1 4.9 25.4 n n n 5.1% 5.0% ( ( ( 33.9% harvested harvested Lo’i being

± ± ± Habitat categories in taro lo’i lo’i in taro Habitat categories

= 103) = 103) = 103) 10.1 10.2 19.2 8.2% n n n 10.2% 62.0% APPENDIX 23 ( ( ( Medium to dense growth dense growth

± ± ± 4.0 3.7 = 119) = 119) = 119) 14.3 4.8% 4.1% n n n 27.3% or mature mature or ( ( ( Early growth Early growth

± ± ± ) of vegetated sample plots by taro hab n from 3 March to 12 November 2004. = 22) = 22) = 22) 3.1 1.5 12.4 n n n 1.8% 1.7% planted ( ( ( Recently Recently 16.3% SD; ±

(at 15 cm) Green cover (canopy) (surface) Green cover Green cover Percent green cover (mean Hawaiian coots at Hanalei NWR

154

APPENDIX 24

List of waterbird species observed during scan surveys at Hanalei NWR during the 2004 field season.

Family Species name Common name Anatidae Branta bernicla Black brant Anatidae Branta sandvicensis Hawaiian goose Anatidae Anas penelope Eurasian wigeon Anatidae Anas americana American wigeon Anatidae Anas platyrhynchos Mallard Anatidae Anas wylvilliana Hawaiian duck Anatidae Anas clypeata Northern shoveler Anatidae Anas acuta Northern pintail Anatidae Aythya americana Redhead Anatidae Aythya affinis Lesser scaup Anatidae Bucephala albeola Bufflehead Ardeidae Bubulcus ibis Cattle egret Ardeidae Nycticorax nycticorax Black-crowned night-heron Threskiornithidae Plegadis chihi White-faced ibis Rallidae Gallinula chloropus sandvicensis Hawaiian common moorhen Rallidae Fulica alai Hawaiian coot Charadriidae Pluvialis fulva Pacific golden plover Recurvirostridae Himantopus mexicanus knudseni Hawaiian stilt Scolopacidae Tringa incana Wandering tattler Scolopacidae Limnodromus scolopaceus Long-billed dowitcher Laridae Larus atricilla Laughing gull