M.A.CB5.H3 3443 R.Pdf
UN1VEF.SITY OF HAWAl'1 LIBRARY
POPULATION STRUCTURE OF THE HAWAIIAN TREE FERN CIBOTIUM CHAMISSOI ACROSS INTACT AND DEGRADED FORESTS 0' AHU, HAWAI'I
A THESIS SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF HAWAI'I IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF
MASTER OF ARTS IN GEOGRAPHY (ECOLOGY, EVOLUTION AND CONSERVATION BIOLOGY)
DECEMBER 2007
By Naomi N. Arcand
Thesis Committee:
Lyndon Wester, Chairperson Stacy Jorgensen Tamara Ticktin We certify that we have read this thesis and that, in our opinion, it is satisfactory in scope and quality as a thesis for the degree of Master of Arts in Geography.
HAWN CB5 .H3 no. ~yy 3
THESIS COMMITTEE
ii ACKNOWLEDGEMENTS
I would like to thank the American Association of University Women for funding this research project and the Army Natural Resources staff for their invaluable hours of staff time, site access, and field support. I would like to express much gratitude to my advisor,
Lyndon Wester for his advice, support, and patience throughout my graduate program at
University of Hawai'i-Miinoa. I also have much gratitude to Tamara Ticktin and Stacy
Jorgensen for their guidance in thesis design, data analysis, and revisions. Thanks to the
Ecology, Evolution, and Conservation Biology Program for loaning research equipment.
I would like to acknowledge Kay Lynch, the hapu 'u horticulturalist extraordinaire, for her time and efforts to grow Cibotium, and also for her feedback and endless positive support for native fern research and awareness in Hawai'i. I thank the following staff and management agencies for supporting and facilitating access to research sites: Talbert
Takahama, Reuben Mateo, Betsy Gagne, and the Hawai'i Natural Area Reserve
Commission; Martha Yent and Kahana State Park; Ray Baker, Nellie Sugii, Alvin
Yoshinaga, and Lyon Arboretum; Earl Pawn and the Department of Forestry and Wildlife
Honolulu Watershed Forest Reserve; Joel Lau and Roy Kam at Hawai'i Natural Heritage
Program; and Clyde Imada at the Bishop Museum. I could not have completed my project without the help of numerous field assistants, including Jane Beachy, Aaron
Shiels, Tom Ranker, Kay Lynch, Lara Reynolds, John Delay, Aurora Kagawa, Maggie
Sporck, JeffMikulina, Kapua Kawelo, Joby Rohrer, Steph Joe, Leann Obra, Dominic
Souza, Will Weaver, Vince Costello, Julia Gustine, Where's George?, Susan Ching, Dan
Sailer, and Charlotte Yamane. I also thank the following people for their assistance and
III invaluable input: Jane Beachy, Dieter Mueller-Dombois, LawTen Sack, Matt
McGranaghan, Ev Wingert, Dave Olsen, Tom Giambelluca, Ross Sutherland, Scott
Lynch, Colleen Moore, Susan Beatty, Dan Palmer, Travis Idol, Dave Palumbo, Matt
Keir, Matt Burt, Steve Mosher, Seth Cato, Krista Winger, Lasha Salbosa, Daniel Toibero,
Bob Kinzie, Frank Howarth, Lloyd Loope, Mach Fukada, Pat Conant, Mark Wright,
Peter Follett, Paul Banko, Jan TenBruggencate, Jeff Mikulina, and Nathan Yuen. Finally, thanks to my parents Renee and Dennis Arcand for their continuous support and encouragement to complete this project.
IV TABLE OF CONTENTS
Acknowledgements ...... iii
List of Tables ...... v iii
List of Figures ...... ix
Chapter I: Introduction...... 10
Problem ...... 12
Purpose ...... 15
Background ...... 16
Ecological Role of Tree Ferns ...... 16
The Genus Cibotium ...... 19
Framing Research in Restoration ...... 24
Disciplinary Framework ...... 27
Significance ...... 28
Research Questions ...... 29
Chapter 2: Methods ...... 31
Locating Study Sites ...... 31
Individual Site Descriptions ...... 34
Kahanahiiiki and Pahole ...... 34
Three Points ...... 37
'Ohikilolo ...... 38
Kahuku ...... 39
Kahana ...... 41
'Aiea Ridge ...... 42 v Lyon Arboretum ...... 43
Plot Measurements ...... 44
Cibotium Distribution and Habitat ...... 47
Outplanting Trial...... , ...... 47
Statistical Analyses ...... 49
Chapter 3: Results ...... 52
Population Structure and Morphological Variation ...... 52
All Measured C. chamissoi ...... 52
Paired FencedlUnfenced Plots ...... 53
All Unfenced Plots ...... 62
C. chamissoi Environmental Recruitment Correlations ...... 64
All Measured C. chamissoi ...... 64
All Unfenced Plots...... 68
Variance in C. chamissoi Recruitment and Abundance ...... 72
Presence of Epiphytes ...... 77
Cibotium Distribution Observations and Habitat Parameters ...... 79
Outplanting Trial...... 81
Chapter 4: Discussion ...... 82
Effects of Ungulate Predation ...... 82
Effects of Weed Control...... 84
Effects of Environmental Conditions...... 85
Morphological Variation ...... 90
Epiphytes...... 92
VI Cibotium Distribution...... 93
Outplanting Trial...... 94
Future Research ...... 95
Conservation Implications ...... 97
Appendix I: Field Data Sheets ...... 99
Appendix II: Outplanting Trial: Individual C. chamissoi Data ...... 103
Appendix Ill: Results of Soil Analyses: Paired Plots ...... 104
Appendix IV: Results of Soil Analyses: All Unfenced Plots ...... 105
References...... 106
VII LIST OF TABLES
1. Measured Cibotium growth rates ...... 22
2. Summary description of research plots ...... 34
3. All plots: average size class morphology ...... , ...... " .... 53
4. ANOV A results: fenced/unfenced immature
C. chamissoi relative abundance ...... 56
5. ANOVA results: weed control and morphological variation ...... 58
6. Two-way ANOV A results: fencing/weed control and trunk length ...... 59
7. Mean differences in morphology: fenced and unfenced plots ...... 61
8. Correlates: All Unfenced C. chamissoi ...... 69
9. Correlates: All Unfenced Plots ...... 71
10. Principal components analysis total variance ...... 73
II. Varimax rotation of three factor solution for variance:
pattern/structure for coefficients...... 75
12. Pearson correlates: multiple regression model variables ...... 76
13. Multiple regression results ...... ' ...... 76
14. MUltiple regression model predictor variables ...... 77
V111 LIST OF FIGURES
Figure Page
1. Location of research plots: O'ahu, Hawai'i ...... 33
2. Kahanahaiki and Pahole plot locations ...... 35
3. Three Points plot locations ...... 37
4. 'Ohikilolo plot locations ...... 39
5. Kahuku plot location ...... 40
6.. Kahana plot location ...... 41
7. 'Aiea Ridge plot location ...... 43
8. Lyon Arboretum plot location ...... 44
9. Population structure and abundance: fenced and unfenced ...... 54
10. Population structure by percentage: fenced and unfenced ...... 55
II. Average population structure and abundance:
fenced and unfenced ...... 56
12. Two-way ANOVA: mean trunk length by fencing and weeding ...... 59
13. Population structure and abundance: all unfenced ...... 63
14. Population structure by percentage: all unfenced...... 64
15. Correlation between trunk length and rainfall:
all C. chamissoi ...... 65
16. Correlation between understory cover
(excluding C. chamissot) and size class ...... 67
17. Correlation between invasive understory cover and trunk length ...... 68
IX LIST OF FIGURES (CONTINUED)
Figure
18. Principal components analysis screeplot ...... 74
19. Correlation between presence of epiphytes
and basal trunk circumference...... 78
20. Correlation between presence of epiphytes and trunk length ...... 79
21. Observations of Cibotium on O'ahu ...... 80
x CHAPTER 1. LNTRODUCTION
The fern forest has an inimitable charm and a distinctive beauty. In architecture and atmosphere, it is unique ... Her radiantfern groves will long remain among Hawaii's noblest treasures. -(MacCaughey 1916)
Hawai'i is considered an ideal location for many types of evolutionary, biogeographical, and ecological studies due to high levels of endemism, isolation, and topographic diversity (Wagner et al. 1985; Kaneshiro 1989; Loope and Mueller-Dombois 1989; Funk and Wagner 1995; Sakai et al. 2002). Hawai'i's montane forests are imperative to surface and ground water supply (Giambelluca 1983; Giambelluca et al. 1986) and concern for their health and functionality has led to conservation efforts as early as the mid 1800's
(Cuddihy and Stone 1990). Tropical island montane forests throughout the world are important in maintaining soil stability, water quality, and at higher elevations are known to harbor natural refugia for rare and endemic species (Sakai et al. 2002).
Tree ferns are an ecologically significant component ofHawai'i's montane forests, and
Hawai'i is home to four distinct endemic tree fern species ofthe genus Cibotium (Palmer
1994), (Cibotiaceae Korall). Known locally as hapu 'u, they are thought of as keystone species (Durand and Goldstein 2001; Mueller-Dombois 2005; Mueller-Dombois in press). Evidence that Cibotium plays an integral role in the community structure and function of Hawaiian forests is substantiated by the large number of native vascular plants found growing on tree fern trunks as epiphytes (Mueller-Dombois et al. 1981;
11 Medeiros e/ at. 1993), the structural role tree ferns occupy in maintaining intact native forest by discouraging invasive species establishment (Buck 1982), and the ability of tree fern fronds to sequester high levels of important nutrients such as nitrogen and phosphorous (Vitousek et al. 1995; Scowcroft 1997). In this regard, Cibotium can be seen as key megaflora of Hawai'i's rainforests. The purpose of this biogeographical study is to examine the population structure, abundance, and vegetation ecology of Cibotium chamissoi Kaulf. across a diversity offorest communities on the island ofO'ahu.
Problem
Unfortunately island ecosystems worldwide are undergoing extreme degradation; human land-use alterations and an influx of non-native species-such as the invasion of alien vegetation and browsing feral goats and pigs-are changing the species composition and ecosystem functions of these centers of endemism (for example, see Loope and Mueller
Dombois 1989; Schofield 1989; Vitousek and Walker 1989; Vtorov 1993; Asner and
Vitousek 2005). Isolated high tropical oceanic islands are considered more vulnerable to alteration by biological invasions due to their inherently high rates of endemic species that evolved in the absence of humans and grazing mammals (Carlquist 1974; Mueller
Dombois 1981; Wagner et at. 1985; Vitousek 1988; D'Antonio and Dudley 1995; Sakai et at. 2002). Ground disturbance by feral pigs is known to facilitate alien plant invasion, and a lesser endemic plant adaptability for such disturbance further enables the success of alien establishment (Mueller-DomboisI981; Vitousek 1988). Invasive plant species can also out-compete native vegetation for nutrients, light, and moisture (for example, see
Mueller-Dombois 1973; Vitousek 1992; Ostertag and Verville 2002; Callaway and
12 Maron 2006). An elevational assessment of plant invasions in Hawai'i found that even at , high elevations the sampled plant communities contained several alien species-whose spread was often facilitated by feral ungulates-and alien species were also present in protected fenced areas (Daehler 2005). The invasive overstory tree Morellafaya (Ait.)
Wilbur and understory invasive ginger Hedychium gardnerianum Shepard ex Ker-Gawl have been shown to alter both forest canopy and understory chemistry dynamics and water content in Hawai'i (Asner and Vitousek 2005). It is of special concern that a species of non-native tree fern, Sphaeropteris cooperi (Hook. ex F. Muel!.) R. M. Tryon grows faster and becomes reproductive earlier than native tree ferns. It has escaped cultivation, and is now invading forests on several of the Hawaiian Islands (Medeiros et al. 1992; Durand and Goldstein 2001).
Feral goats and pigs on islands over-browse native vegetation, suppress native vegetation regeneration, facilitate weed dispersal, and exacerbate erosion (Spatz and Mueller-
Dombois 1973; Smith 1985; Cuddihy and Stone 1990; Stone et al. 1992). Feral pigs also dig up large areas for mud wallows and forage, significantly altering the structure of the native understory (Spatz and Mueller-Dombois 1973; Mueller-Dombois et al. 1981;
Smith 1985; Cuddihy and Stone 1990; Stone et al. 1992) and are known to target the starchy cores of Cibotium trunks as a major food source (Diong 1982). Pig herbivory of . the trunk starch can cause mortality of the tree fern, and always does so if the fern growth apex is within reach (Mueller-Dombois et a/. 1981). Tree fern starch was found to compose over 90 percent of pig' stomach contents in Kilauea forest on Hawai'i (Mueller-
13 Dombois et al. 1981) and was found to compose the bulk of their diet in KIpahulu Valley on Maui (Diong 1982).
Casual observations of Cibotium abundance and size in comparison to the island of
Hawai'i may indicate a decline on O'ahu (Palmer 2003). To date, reason for the reduced abundance and size of Cibolium on O'ahu has been a subject of speculation. Possible explanations include high browsing pressure from feral pigs, displacement by competitive alien vegetation, illegal collection of tree fern trunks for the orchid industry and landscaping (Dr. Dieter Mueller-Dombois, personal comm.; Dr. Lyndon Wester, personal comm.), and high rates of predation by an introduced two-spotted leafhopper
(Jones et at. 2000; Palmer 2003). Alternatively, the current populations of Cibotium on
O'ahu may reflect a successional process or be under topographic control. Differences in
Cibotium glaucum (Sm.) Hook. & Am. size and abundance have especially been noted between the islands of Hawai'i and O'ahu. On Hawai'i, C. glaucum reaches heights over
20 ft. tall (Nelson and Homibrook 1962; Wick and Hashimoto 1971) and achieves extremely high densities with over 2,000, or as many as 5,000 mature individuals per hectare (Mueller-Dombois et at. 1981; Drake and Mueller-Dombois 1993) where it composes 77.3 percent cover in the subcanopy of Metrosideros polymorpha Gaudich. forest (Mueller-Dombois et al. 1981). Substrate age, drainage, and topographic differences between the islands have been considered as possible explanations for the difference in size and abundance of C. glaucum. It has been suggested that as nutrient concentrations vary with age of substrate, the younger soils, taller forest structure, and larger areas of a more flat topography on the island of Hawai 'i favor C. glaucum (Kapua
14 Kawelo, personal comm.; Matthew Keir, personal comm.; Joel Lau, personal comm.).
Another possible cause for observed differences could include as yet unidentified fungal pathogens. In short, the current population status of Cibotium is unknown on O'ahu and there is a lack of knowledge regarding abundance and distribution of Cibotium across the main Hawaiian Islands.
Purpose
Because the population status of Cibotium on O'ahu is currently unknown, this biogeographical study examines the population structure, abundance, and potential restoration importance of the species Cibotium chamissoi across a spectrum of "intact" and "degraded" forest communities in the Wai'anae and Ko'olau Mountains ofO'ahu.
The terms intact and degraded are inherently subjective, and for the purposes of this study are defined and used as follows: "intact" forests are considered to be constituted mainly of native species and are presumed to function as a healthy ecological system; conversely "degraded" forests are those heavily invaded by alien species and/or feral pigs, with low cover of native species and possibly reduced native species diversity. The island of 0' ahu is the most developed of the Hawaiian Islands, and has undergone intense development and subsequent forest degradation. Intact native forests are generally confine'd to the upper elevations. Feral ungulates and alien weed species have degraded lower forests and even pockets of higher elevation forests. Therefore the mid-elevation zone is an ideal area to study the tropical island forest landscape because here is found a spectrum of intact native and degraded alien habitat where Cibotium chamissoi is the common tree fern species. It is accepted that Cibotium are truly an integral part of
15 Hawaiian forest communities, therefore research is needed to determine if they are indeed
disappearing from O'ahu's forests. Though this study will not assess all of the potential
causes for a possible Cibotium decline, we may be able to determine baseline patterns in
abundance and population structure o(c. chamissoi. By isolating the impacts of feral
ungulates and examining the effects of environmental gradients on C. chamissoi
populations on O'ahu, at the very least, we may be able to identifY or dismiss several
possible explanations for variations in C. chamissoi recruitment and abundance.
Background:
Ecological Role of Tree Ferns
Tree ferns form a major component of most south temperate and tropical rainforests
(Walker and Aplet 1994) and are a conspicuous member of the world's wet tropical montane forests (Conant et af. 1994). An important early observation of global tree fern distribution by MacCaughey (1916) noted the wide range of temperatures in which tree ferns occur, but a consistency in their preference for areas of high atmospheric humidity.
Tree ferns as a group have been recently reclassified based upon molecular evidence to be placed together in the order Cyatheales (A. R. Smith et af. 2006). Tree ferns are globally distributed in tropical, subtropical, and south-temperate regions (Gastony 1981) and have been divided into eight families, 15 genera, and 663+ species (A. R Smith et af.
2006).
Most but not all tree ferns have a tall, trunk-like rhizome and tree-like growth form
(Korall et af. 2006). Structurally, tree ferns form a major subcanopy layer in tropical
16 forests and can be seen as a major gatekeeper of shade and moisture in the forest understory. Their large fronds protect soil layers from erosion during heavy rainfall events and possibly increase water percolation into soil, as they may divert a significant amount of water away from their central stem (Mueller-Dombois 2005). Tree fern trunks also provide a magnificent substrate for epiphytes and otherwise terrestrial vascular plants (Medeiros et af. 1993; Mueller-Dombois 2005). The trunks allow vascular plants a competitive advantage in establishing slightly above the forest floor, where light is not as limited. It is the tree fern trunks, with their moisture-holding adventitious root mass, which makes the growth substrate attractive to orchid growers: tree ferns are often unsustainably harvested for the orchid industry (Conant et aZ. 1994). Regeneration of woody vegetation in a cloud forest on Maui was found to be correlated to the abundance of moss-covered coarse woody debris, and especially crucial to regeneration of the understory (Santiago 2000). As older tree fern trunks are often covered in mosses, they are a preferential substrate for seedling establishment. It was also noted that Cibotium chamissoi was one of few species-and the most abundant species-to grow directly from the mineral soil, whereas most other species commonly start as seedlings on logs above the soil surface (Cooray 1974; Santiago 2000; Mueller-Dombois 2005; Mueller
Dombois in press). A comparison of epiphytes on the native Cibotium to the invasive
Sphaeropteris cooperi in KIpahulu Valley, Maui found epiphytes to be more abundant on
Cibotium trunks (Medeiros et aZ. 1993). Possible explanations included the age of
Cibotium trunks, which were much older than the fast-growing S. cooperi, and the possible mechanical/chemical resistance of S. cooperi trunks to epiphytic establishment
(Medeiros et af. 1993). The retention of dead fronds around the trunk as tree fern "skirts"
17 is a trait exhibited by several species across the world and may offer resistance to climbers and large epiphytic plants that would otherwise reach the crown and prevent the growth of new emerging fronds (Page and Brownsey 1986). It is noted however that epiphytes often establish below the skirt (Page and Brownsey 1986). Cibotium chamissoi is described as retaining a persistent skirt of dead fronds (Palmer 1994).
Ferns in general are thought to have an important influence in regeneration processes and early forest successional stages (Schmitt and Windisch 2006), and tree ferns that share similar sub-canopy growth habits to Cibotium have been studied in other tropical forest systems. Important observations were reported on the different edgelinterior growth habits and survival of three species (Alsophilafirma (Baker) D. S. Conant, Lophosoria quadripinnata (Gmel.) C. Chr., and Sphaeropteris horrida (Liebm.) R.M. Tryon) in a cloud forest of Mexico, with results indicating Lophosoria is an early successional pioneer of disturbed areas whereas Sphaeropteris is more successful within the forest interior (Bernabe et al. 1999). A study six tree fern species in Costa Rica were also found to display habitat preferences as measured by growth rates and age (Bittner and Breckle
1995). Ash (1987) reported improvements in methods for estimating age for the tree fern
Cyathea hornei (Baker) Copel. in Fiji using analysis ofleaf scar spacing, and Schmitt and
Windisch (2006) determined frond production and phenology of spore release for
Alsophila setosa Kaulf., noting the need for baseline growth and reproduction data due to the threats to local tree fern populations in southern Brazil, including commercial exploitation for the orchid industry and landscaping. Indeed, Cibotium barometz (L.) J.
Sm. has been listed in Appendix II of the Convention on International Trade in
18 Endangered Species (CITES) due to its harvest for the Chinese herb medicine trade and the ornamental plant trade (Zhang et al. 2002) along with other tree ferns in the genera
Cyathea and populations of Dicksonia from the Americas. Dicksonia antarctica L'Her. in southern Australia is also commercially valuable, and has been successfully propagated under a tree plantation to supply the horticultural trade (Vulcz et a1. 2002). Other impacts to tree fern populations have been noted from logging activities in Australia, where younger forests regenerating from clear-cut harvesting had significantly less abundance of tree ferns than older succession forests (Lindenmayer et af. 2000) and significantly lower survival rates after clear-cuts (Ough and Murphy 2004). Tree ferns were also found to be more abundant in forests regenerating after wildfire than in clear-cut areas in the same stage of development, suggesting that clear-cut harvesting alters natural forest regeneration (Ough 2001), rather than implying that tree ferns are not important in early successIOn.
The Genus Cibotium
In the main Hawaiian Islands, the dominant tree fern is of the genus Cibotium. A recent study by Korall et al. (2006) has determined the Dicksoniaceae family to be non monophyletic, and Cibotium has been reclassified to its own new family of Cibotiaceae as described by Korall (Smith et al. 2006). Korall et al. (2006) recognize the single genus of Cibotium and 11 species within Cibotiaceae.
Palmer (1994) has studied the Hawaiian Cibotium in detail and the following information is derived from his work. The genus is distributed from Papua New Guinea, China, and
19 Assam in the Old World, to Mexico and Central America in the New World, and
Hawai'i. The island of O'ahu is home to three distinct endemic species of Cibotium: C. glaucum, C. menziesii Hook., and C. chamissoi. Cibotium chamissoi is distinguished by its frequent retention of a dead skirt of fronds, and restriction of golden or mustard colored hairs to the base of the stipe and apex of the caudex. Cibotium glaucum may sometimes be confused with C. chamissoi, but is distinguishable by the light bluish-gray or bluish-white waxy coating on the undersides of the frond blade. Cibotium menziesii is easily identified by the straight, dark hairs that cover the entire length of the stipe.
Cibotium chamissoi is found on all the major Hawaiian Islands save Kaua'i, where it has not been collected. It is the most common lower elevation Cibotium on O'ahu, and can be found as low as 50 m in protected areas of shady moist habitat. It is commonly seen up to
600 m, and at higher elevations is then joined by C. menziesii and C. glaucum. Cibotium glaucum is the most dominant species at high elevations. Cibotium chamissoi is determined as the appropriate subject of this study due to its common distribution on
0' ahu in the elevational range that overlaps with intact native forest and disturbed forest areas.
Although Cibotium spores are mainly dispersed by wind, successful establishment and development of the gametophyte is largely dependant on the suitability of microhabitat.
Both the gametophyte and sporophyte of C. glaucum have been classified as shade requiring plants, demonstrating restricted growth and photobleaching in full sunlight
(Friend 1974). However, mature pteridophytes often drop spores in the neighborhood of parent plants (Sheffield 1996), so close in fact that Conant (1978) documented Cyathea
20 arborea, a tree fern of8 m tall, to drop the majority of its spores within 7.5 m of the parent.
Cibotium is characterized by unique spore morphology among the other families of
Cyatheales, described as globose-tetrahedral with prominent angles and a prominent equatorial ridge (Gastony 1982; Smith et al. 2006). Sixty-four spores are produced per sporangium (Gastony 1982) and an average fertile frond of Cibotium chamissoi produces
700,000,000 spores (Dyer 1979). Gastony (1982) published a detailed account and scanning electron micrographs of spore morphology for Cibotium species from Hawai' i,
Central America and Mexico, and Southeast Asia. These detailed images of spores will be useful in any future studies of soil spore banks to understand current and historical reproductive potential of Cibotium. A chromosome base number ofK = 68 is also unique to Cibotium amongst the other tree fern genera (Smith et al. 2006).
Other research on Cibotium in Hawai'i has been focused on general distribution and elevational ranges of pteridophytes (MacCaughey 1918) and individual growth rates
(Ripperton 1924; Wick and Hashimoto 1971; Walker and Aplet 1994), and growth rate and reproductive comparisons with introduced tree ferns (Durand and Goldstein 2001).
Growth rates for Cibotium are summarized in Table 1.
21 Table 1.
Measured Cibotium Growth Rates
Author Species Island CIIlIYr Durand and Goldstein (2001) C. chamissoi Oahu 3.00 Walker and Aplet (1994) C.l!.laucum Hawaii 4.44 - 6.48 Wick and Hashimoto (1971) C. glaucum Hawaii < 5.08 Ripperton (1924) C. charnissoi Hawaii 11.05* * Overeshmate based upon flawed methodology; results reported at 4.31 mches/yr
Native Hawaiians used the starchy trunk pith of Cibotium as a source of food in times of famine and to feed their pigs (Ripperton 1924; Fosberg 1942; Palmer 2003). The young fronds were cooked and eaten with taro and meat as a source of greenery, and the pulu, or golden hairs that cover new emerging fronds of Cibotium were also used for dressing wounds and embalming the dead (Fosberg 1942). In the late 1800's, a high volume of
Cibotium pulu was exported from Hawai'i to the U.S. mainland for filling mattresses and pillows, but the industry died because pulu was not durable enough (Fosberg 1942;
Cuddihy and Stone 1990; Palmer 2003). Tree fern starch, mostly from C. chamissoi, was locally marketed as a food and detergent in the 1920's, and a skilled workman was thought to be able to harvest approximately 1,000 pounds of starch per day, or 20 large tree ferns (Ripperton 1924). "Fortunately for those who enjoy the sight of a forest of gracefullehua with an understory of tree-ferns, the venture soon failed," (Fosberg 1942:
17). An additional U.S. Forest Service report attempted to estimate total volume of tree ferns for supply of a current and future commercial starch and orchid medium industry
(Nelson and Homibrook 1962).
22 The two-spotted leafhopper, Sophonia rufofascia, although originally first documented on O'ahu in 1987 (Heu and Kumashiro 1989), quickly spread within five years to all major neighbor islands (Fukada 1996; Culliney 1998). Though Jones et al. (2000) found
S. rufofasica feeding to cause necrosis of uluhe fern (Dicranopteris linearis Burman) fronds, their experiment was perhaps useful to identify effects of leafhopper feeding under a high density of constant feeding pressure rather than a more realistic situation of lower leafhopper densities that are able to move from plant to plant in a given area.
Sophonia rufofasica has been implicated in the observed large patches of uluhe dieback, but evidence is lacking to conclude that the leafhopper is solely responsible (Johnson et al. 2001). Jones et al. (2000) also observed leafhopper oviposition to disrupt vascular bundle cells in Cibotium fronds and often this was observed to cause mortality of segments of the pinnule below the egg, or in other words tips of the segments of the fern frond rather than the entire fern frond. Because the xylem and phloem are both disrupted, drought stress may further facilitate the effects of leafhopper oviposition on frond pinnule mortality (Jones et al. 2000). Invasion of native forests by the exotic weed tree Morella faya has been shown to be related to two-spotted leafhopper densities and Mfaya itself is identified as a preferential host for oviposition when compared to the native dominant tree Metrosideros polymorpha, but M faya invaded forest areas are thought to be "a more favorable habitat for the two-spotted leafhopper" with spillover effect to other native forest vegetation (Alyokhin et al. 2004: 6). It is of concern that a study of two-spotted leafhopper egg density (eggs per square meter of leaf surface area) on nine plant species across multiple islands found egg density to be higher in closed wet natural habitats
(Alyokhin, et at. 2001), which is the realm ofthe tree fern. Not surprisingly then, 23 Alyokhin et al. (200 I) found Cibotium chamissoi to harbor a significantly higher density of two-spotted leafhopper eggs across sites surveyed. However, the authors state
" ... overall egg density per unit ofleaf area was fairly low. Therefore, even though leafhopper oviposition may kill the distal leaf tissue of some hosts (Culliney 1998; Jones et al. 2000), it is unlikely that oviposition alone can cause significant damage to affected plants" (Alyokhin et al. 2001: 667). It is encouraging to note that Sophonia rufofascia eggs have parasitoids (Johnson ef al. 2001), resulting in an average of 40% of eggs parasitized across all species and all sites in one study (Alyokhin et al. 200 I). This analysis of the literature on two-spotted leafhopper impacts to Cibotium reveals that despite feeding and oviposition damage to frond pinnules on multiple Hawaiian Islands, it is highly unlikely that Sophonia rufofascia alone has caused a major decline in abundance or size of Cibotium on O'ahu.
Framing Research in Restoration
Tree ferns are thought to be key species of the forest understory, yet in forests degraded by feral ungulates and invasive species, little is known about the potential usefulness of tree ferns in forest restoration. However, a few attempts have been made that lay the groundwork for future restoration efforts. An admirable dissertation by Becker (\976) studied the morphological variation and abundance of Cibotium in a variety of undisturbed forest communities on the island ofHawai'i. Becker also conducted a transplant trial into a lOx 6 m fenced area on open pasture "almost without any shading" that once supported Acacia koa GraylMetrosideros polymorpha forest on Hawai'i: ten C. glaucum and two C. chamissoi were felled, all live fronds were cut off, and they were
24 placed horizontally along the ground on top of the kikuyu grass (Pennisetum clandestinum Hochst. ex Chiov.) in June, 1973. Six of the trunks were covered with the cut-off fronds to protect the trunks from the direct sun rays. Some of the trunks had lateral buds (17 total buds) without expanded fronds. The fence lasted 16 months. In this time, II of the trunks produced new fronds, and all trunks produced new adventitious roots. The grass under the trunks died, and the roots had penetrated into the soil. Trunks covered by the dead fronds produced more roots than the exposed trunks, and only one of the total trunk buds produced fronds. Before the number and length of fronds produced could be measured, cows broke into the exclosure and ate all of the new fronds.
Therefore survival cannot be concluded.
Ripperton (1924) tried re-establishment of Cibotium on Hawai'i by planting crowns and lateral shoots in a pastureland adjacent to forest containing tree ferns. The experiment was discontinued early, but the goal was "to determine whether the tree ferns could be successfully planted on areas that had been denuded of their original growth and on which there was no shade," (p. 5). He found that the crows and lateral shoots started growing during the rainy winter months but all died during the summer. Results from two other experimental sites within the forest interior showed all crowns and lateral shoots became successfully established. Ripperton never stated how long he monitored the transplants, but it is assumed to be one year or less, and therefore long term survivorship cannot be concluded with certainty.
25 The regeneration of forest species with ungulate exclusion has been studied and reviewed
(for example, see Spatz and Mueller-Dombois 1973; Loope and Scowcroft 1985; Stone et al. 1992; Cabin et al. 2000). Loope and Scowcroft (1985) acknowledge that few exc10sures exist in the Metrosideros rain forest zone across the state of Hawai'i (realm of the tree fern), but their study of vegetation responses within other exclosures show a general trend of native vegetation maintaining itself or increasing with the exclusion of ungulates. However, their review concludes that native vegetation recovery was limited if degradation continues through invasion of alien vegetation (Loope and Scowcroft 1985).
This is concurrent with findings by Cabin et al. (2000) in their dry forest regeneration study.
Dissertation research by Loh (2004) on the experimental removal of the weedy tree
Morella faya on Hawai'i revealed the ability of Cibotium glaucum, other ferns, and
Metrosideros polymorpha to regenerate in a rather limited capacity (after three years) after M faya individuals were selectively girdled in incremental stages and left standing to die slowly. Unfortunately however, Loh found the overall abundance of native species establishment in the girdled plots to be five to ten times lower than alien species recruitment (Loh 2004). Partial restoration of the native forest is expected in the girdled control plots of M faya, but without follow-up weed management the possibility for re invasion or new invasive species to establish is high (Loh 2004).
Mueller-Dombois (2005; in press) has generated some intriguing suggestions for use of
Ctbottum chamissoi in restoration on O'ahu that have yet to be tried. In his discussion of
26 Cibotium and their importance to the ecosystem as key species, he notes "they are those whose population dynamics [have 1a strong effect on the other species in the community"
(Mueller-Dombois 2005: 57). Hapu 'u can be seen as microhabitat enhancers, and in
"using a silvicultural approach to forest restoration," special consideration should be given to using tree fern trunks as native seed germination sites (Mueller-Dombois 2005:
58). In addition, selective de-limbing of alien trees and fencing in a "klpuka" configuration are suggested for restoration strategies (Mueller-Dombois 2005; Mueller-
Dombois in press). If tree ferns would be expected within an area and are lacking,
Mueller"Dombois (2005; in press) suggests their use in restoration reintroduction plantings, within fences are several: they can be planted directly into the soil, they may facilitate water percolation into the understory soil by forming a protective subcanopy layer in heavy rainfall events, and their trunks provide an ideal substrate for epiphytic native plant establishment. In order to test these ideas of forest restoration using Cibotium chamissoi on O'ahu (a goal of my dissertation research project), it is first necessary to gather baseline data on current C. chamissoi population structure and comparisons of its population structure across intact and degraded forest communities.
Disciplinary Framework
Geography takes a multi-disciplinary approach to studying the relationships among natural processes, areas, society, culture, and the "interdependence of all these over space," (Christopherson 2004: 2). The field of biogeography is generally concerned with the geographical distribution of plant and animal populations, and the relationship of human culture to the environment (Livingston 1992). Brown and Lomolino (1998) define 27 biogeography as the "science that attempts to document and understand spatial patterns of
biodiversity," (3). The spatial and temporal scales oflandscape change, resource
management, and the biogeographical emphasis on Cibotium population structure and
distribution all justify the relevance of this study to the discipline of geography.
Biogeography is a useful sub-discipline through which tree fern population dynamics can
be examined across a suite of microhabitats and across the larger montane landscape of
O'ahu. Scale of this analysis will vary from individual C. chamissoi morphology, to
population-level size class and recruitment, to vegetation community diversity, to the
assessment of environmental controls such as slope, elevation, and rainfall. These
measures will also be put into the temporal context of historical management regimes and
current evidence of habitat disturbance. The human/environment interface, a long
tradition in biogeography, is largely the motivator of this study: to inform forest management decisions for conservation purposes.
Significance
Because Cibotium are keystone species in Hawaiian forest communities, we may think of their absence as indicators of declining forest health. Indeed, ferns in general are considered an integral part of most Pacific island forest ecosystems. Research that determines current population structure of Cibotium in its native range on O'ahu will provide a basic understanding of spatial and temporal dynamics. It is hoped this baseline data will help guide future understory restoration efforts. It is possible that this study may help to inform restoration efforts in other degraded Pacific island forests with a native
28 tree fern understory structure. To date there is a lack of knowledge of the status of tree ferns in degraded forest communities. Unfortunately however, tree ferns across the world are showing signs of stress. An assessment by Wardlaw (2002) of the 1997 IUCN Red
List afThreatened Plants concluded that 206 tree fern species (or 31% of the total species analyzed) are threatened. While ex situ conservation of threatened tree ferns is a useful just-in-case strategy in preventing species extinction, Wardlaw (2002) writes "ideally, conservation should be in situ and involve maintaining the ecosystem that contains the species" (396). In the case of Clbatium chamissal, land managers would benefit from conclusive scientific data that assists in them in deciding how to prioritize resource allocation in forest management and restoration. In other words, forest managers may ask the questions, "should we fence or weed? Or are both necessary?"
Research Questions
To frame the goals of this research project in an ultimate context, the question is whether
Clbatlum should be used in restoration efforts to facilitate the natural regeneration of other native species and to discourage invasive species establishment. The answer to this question would require a long-term study, perhaps over decades, but the foundation can be laid here. This project will address the following subset of research questions:
I) What are differences in C. chamissai population structure and morphology between paired fenced and unfenced plots in the Wai'anae Mountains, and between all unfenced plots in both the Wai'anae and Ko'olau Mountains?
29 2) Which environmental factors are correlated to C. chamissoi recruitment between sample sites: elevation, annual rainfall, slope, aspect, soil nutrients, higher percentages of native (vs. alien) species, and/or percent canopy cover?
3) Using principal components analysis, which measured variables are most responsible for variance in collected data among research plots: slope, aspect, rainfall, fencing, weeding, native or invasive percent understory cover, native or invasive percent overstory cover, overstory density, soil nutrients, or pig activity? Using multiple regression, which of these variables explain the most variance in C. chamissoi sporeling abundance across all plots? .
4) What is the frequency of epiphytic colonization on Cibotium chamissoi individuals and which vegetation community characteristics, environmental characteristics, and morphological characteristics predict presence or absence of epiphytes?
5) Where has Cibotium been recorded to occur on O'ahu by Army Natural Resources and
Hawai'i Natural Heritage Program?
6) What is the survivorship of 15 nursery grown and experimentally outplanted C. chamissoi into Kahanahaiki, a fenced area of forest restoration habitat in the Wai 'anae
Mountains where other C. chamissoi naturally occur?
30 CHAPTER 2. METHODS
Locating Study Sites
The island ofO'ahu is ofvo1canic origin ca. 3.7 million years old, located 22 degrees north of the equator in the Hawaiian Archipelago, and is 1,574 km 2 in area (Wagner et al.
1990). The Ko'olau Mountains shape the windward eastern side of the island, where the trade winds predominantly cause higher levels of rainfall. Maximum annual rainfall at the summit of the Ko'olau range reaches 7000 mm, and 1500 mrn along the windward coast
(Giambelluca et ai. 1986). The Wai'anae Mountains form the leeward western side of the island where rainfall is typically much lower, especially along the coast. However, Mt.
Ka'ala in the Wai'anae Mountains is the highest mountain on O'ahu at 1,225 m, and it receives upwards of 2,000 mm annual rainfall (Giambelluca et al. 1986).
In order to address research questions one through four, a total of sixteen lOx 20 m research plots were established and surveyed from August 2005 to May 2006. Twelve research plots were established in the Wai'anae Mountains and four in the Ko'olau
Mountains. Research site selection utilized areas of accessible known Cibotium chamissoi groves based upon personal observations and interviews with land managers and botanists. Specifically, because the highest priority of this study was to measure
Cibotium chamissoi population structure, plots were selectively located within the research site area in order to encompass at least 15 individuals. Sites were selected in the
Wai' anae Mountains for pairing plots inside and outside of fenced ungulate exclosures, and the paired plots were located with similar elevation, slope, aspect, and vegetation 31 composition to isolate the effects of browsing goats and pigs on C. chamissoi population structure. Often finding groups of C. chamissoi inside of fenced areas was relatively easy, whereas locating areas that contained at least 15 individuals outside of the fence was extremely difficult and at times not possible. A search for at least two hours was conducted for C. chamissoi groves and the largest numbers of individuals were included wherever possible. The paired plots were restricted to the Wai'anae Mountains since the majority of ungulate exclosures are in this area. A survey of the single ungulate exclosure in the Ko'olau Mountains (managed by Army Natural Resources) at Pe'ahinai'a along the
. summit reveals a majority of sparsely distributed C. glaucum at the higher elevations
(consistent with the description of elevational distribution by Palmer 2003), and this exclosure was decided to be mostly unsuitable habitat for C. chamissoi. The Wai'anae research sites were located in Kahanahaiki, Pahole Natural Area Reserve, Three Points, and 'Ohikilolo (see Figure I). The four unfenced Ko'olau Mountain research plots were established in an effort to sample across the north-south span of the range, and include
Kahuku, Kahana Valley, 'Aiea Ridge, and Lyon Arboretum (see Figure I). Please refer to
Table 2 for a summary description of each research plot for elevation, aspect, slope, estimated rainfall, date of data collection, and management history. Research permits were obtained as necessary for Kahana State Park, Pahole Natural Area Reserve, and
State of Hawai'i Division of Forestry and Wildlife managed Honolulu Watershed Forest
Reserve and Mokule'ia Forest Reserve. Permission and access were granted for work in
Lyon Arboretum, and for escorted work in U.S. Anny training areas of Kahuku (located in Kahuku Training Area), 'Ohikilolo and Kahanahaiki (both located in Makua Military
Reservation).
32 Figure 1.
Location of Research Plots: O'ahu, Hawai'i
,e Legend o Research Plots (16) -- Annual Rain fall (mm)
Elevation (m)
High : 1233
Low : 0
1:250,000 o 5 10
33 Table 2.
Summary Description of Research Plots
Rain/al Previously Elevation Slope Aspect I(mm Plot Name Fenced Weeded Date Read (m) ( ") (0 ) per vr)* Ohikilol0 . In yes yes 7·Nov-OS 920 20 20 1400 Ohikilolo - Out no yes 17-Aug-05 883 32 10 1400 Ohikilolo - Out weedy no no 16-Aug-05 867 31 340 1400
Kahanahaiki - In yes yes 10-Aug-05 590 7 320 1400 Kahanahaiki - Out no no 27-Soo-05 600 19 320 1400 Kahanahaiki - Out steep no no 27-Aug-05 585 34 340 1400
3Points - In yes yes 14-Feb-06 830 13 20 1600 3Points - Out no no 27-Feb-06 850 22 15 1600
Pahole - Inl yes no 16-May-06 635 34 40 1400 Pahole - Out! no no II-May-06 675 24 50 1400
Pahole - In2 yes no 12-Mav-06 670 34 30 1400 Pabole - Out4 no no 19-May-06 685 37 40 1400
Lyon Arboretum no yes 6-Apr-06 230 29 30 3100 Aiea Ridge no no 8-Apr-06 440 28 160 4100 Kahana Valley no no 12-Apr-06 135 27 95 4000 Kahuku no no 17-Apr-06 337 37 30 2200 • Annual ralOfall amounts esllmated usmg HawaII" Ramfall Atlas (Glambelluca €I al. 1986).
Individual Site Descriptions
Kahanahaiki and Pahole:
The three Kahanahaiki research plots are located within the Kahanahaiki Management
Unit in the Makua Military Reservation along the northeastern rim of Makua Valley (see
Figure I). The area is managed by Anny Natural Resources (ANR), a contracted group of
34 civilian employees within the Environmental Division ofthe Department of Public
Works of the U.S. Army Garrison, Hawai 'i. The four Pahole research plots were placed within the Pahole Natural Area Reserve, which is managed by the state of Hawai ' i and is located along the eastern boundary of Kahanahaiki. Fenceline and plot locations are shown in Figure 2.
Figure 2.
Kahanahaiki and Pahole Plot Locations
0.3 0 0.3 0.6 Mil es N "~' ~~h~.n~.~h.~I~~F.·n ·C.·lln·."~~~~~~~"""""'" W*E ~ paho l e Fence!!ne •••• Future Kupuna Fencellne ·0 Plot GPSPoin t s
35 Kahanahaiki plots were located and surveyed in August and September, 2005. Pahole plots were located and surveyed in May 2006. The Kahanahaiki and Pahole fences were constructed and the enclosures made pig free by 1998 (Army Natural Resources Center
2004).
Kahanahaiki fence encompasses 90 acres, and Pahole gulch fence encompasses 217 acres. The unfenced Pahole plots were located in an area that will be fenced soon-the
Kapuna fenceline was under construction but not enclosed at the time of this study. These plots will enable comparison of tree fern populations before and after fence construction when re-surveyed in future dissertation research. Vegetation throughout the Kahanahaiki and Pahole area is considered mesic forest. Weed control efforts have continued within both fenced areas and in small areas outside of fences. General estimates of amount of weed control are 93 .5 total hours for Kahanahaiki in the "Maile Gulch" area where the inside research plot is located (Jane Beachy, personal comm.). Weed control has heen conducted inside the Pahole fence, though not across the area of the research plots
(Talbert Takahama, personal comm.). Feral pig and goat control has been conducted by
ANR staff outside the fence along the ridge adjacent to the two unfenced Kahanahaiki plots, in addition to the larger area around the Kahanahaiki fence beginning in 1998
(Army Natural Resource Center 2004). A total of 167 pigs and 67 goats have been removed from the area since 1998 (Army Natural Resource Center 2004).
36 Three Points:
The two research plots at the Three Points site are located within MokuH! ' ia Forest
Reserve, bordering the southeast rim of Makua Valley and managed by the state of
Hawai'i (see Figure I). The approximately six acre fenced exclosure surrounds mesic forest and was constructed and the enclosure declared pig free in 200 I. Pig browsing pressure in this area is high, and occasionally small pigs have heen able to get inside the enclosure. At the time of my research, at least two small pigs were inside the fence and had been digging in an area adjacent to the research plot. The plot was intentionally delineated to avoid the area disturbed by pig digging, and hunters were contacted to remove the pigs. See Figure 3 for fenceline and plot locations.
Figure 3.
Three Points Plot Locations
D 3 Points Fenceline o Plot GPS Point
37 Prior to fence construction, 44 pigs were removed from the area by the Department of
Land and Natural Resources in cooperation and with the support of ANR between
January 2000 and October 2000 (Army Natural Resource Center 2004). The number of
feral pigs caught in this area is very high for the time span of only 10 months. Since
Mokule'ia Forest Reserve has been opened to hunters, an additional 15 pig catches from
the area had been reported by September 2003 (Army Natural Resource Center 2004).
Weed control has and continues to be conducted at Three Points: ANR collaborates with
the State for an offsite management partnership and estimate 494 hours of weed control
over four years 2002 - 2006 inside the exclosure (Jane Beachy, personal comm.).
'Ohikilolo :
The three 'Ohikilolo research plots are located within 'Ohikilolo Management Unit along
the southern ridgeline ofMakua Valley in Makua Military Reservation, managed by
ANR (see Figure 1). The vegetation is considered mesic forest. The fenced forest exclosure was built and free of goats by 1999 and encompasses approximately 2.5 acres.
An additional fenceline runs the length of 'Ohikilolo ridge in order to reduce the browsing pressure of feral goats within Miikua Valley from neighboring goat source
populations, and goat populations were noted to be significantly reduced by 2000, although some goats still inhabited lower Makua Valley in greatly reduced numbers until
2004 (Kapua Kawelo, personal comm.). Feral pigs are thought to be absent or rare in the area because of the steep cliffs and slopes below and along the ridge, and ANR has never detected pigs in the area (Army Natural Resource Center 2004). See Figure 4 for exclosure fenceline and plot locations.
38 Figure 4.
'Ohikilolo Plot Locations
, /\/ Ohikilolo Fenceline o Plot GPS Point .*s .
The ANR staff and volunteers have spent roughly 272 hours conducting weed contTol inside the forest exclosure and outside the fence in the " Pferalyxia macrocarpa Gulch" area over four years 2002 - 2006 (Jane Beachy, personal comm.).
Kahuku:
The single Kahuku plot is located within Kahuku Military Training Area, on the northern leeward side of the Ko'olau Mountain range and managed by ANR (see Figure I). The
39 plot is located near a ridge 4WD road that is occasionally used by military vehicles and soldiers for training activities. See Figure 5 for plot location.
Figure 5.
Kahuku Plot Location
, o Plot GPS Point ..• '
Weed control has not been previously conducted within the research area, which is considered to be predominantly native wet forest vegetation with a dominant
Melrosideros pO/y/llO/pha overstory. There is no public hunting in the vicinity of the research site. 40 Kahana
The research plot in Kahana Valley State Park is located at approximately 135 m elevation, close to the valley bottom on the nonhern side of the valley (see Figure 1). The vegetation community is considered to be coastal mesic forest (Wagner ef al. 1990) with a koa (A cacia koa) and hala (Pandanus lecforius Parkinson ex Zucc.) overstory, which was a unique site in this study. No weed control has been conducted within the research area. See Figure 6 for plot locati on.
Figure 6.
Kahana Plot Location
0.5 o 0.5
Miles o Plot GPS Point
41 According to Hawai'i State Division of Forestry and Wildlife records, public hunters reported a total of 20 feral pigs harvested in Kahana Valley over the 10 month period of this study.
'Aiea Ridge:
The 'Aiea Ridge plot location is within the State ofHawai'i managed Honolulu
Watershed Forest Reserve (see Figure I). Located near the main ridge hiking trail in a less steep area, weeds appear to be dispersed along the trail and were observed to infiltrate the plot area. There were also signs of disturbance caused by feral pigs. lndeed, hunters with dogs were in the vicinity as we surveyed the research plot. No weed management has been conducted within the area. See Figure 7 for research plot location.
42 Figure 7.
'Aiea Ridge Plot Location
0.5 o 0.5
Miles o Plot GPS Point s
Lyon Arboretum:
The Lyon Arboretum is managed by the University of Hawai'i in Manoa Valley (see
Figure I). Although not fenced, feral pig control has been administered in the Arboretum through hunting. The plot was located in "Valley 4C" within a dense C. chamissoi grove.
Weed management was conducted over roughly three days in the area of the research plot in 1981 (Ray Baker, personal comm.). See Figure 8 for plot location.
43 Figure 8.
Lyon Arboretum Plot Location
0.5 o 0.5
Miles o Plot GPS Point s
Plot Measurements
Plots were run 20 m down-slope along the same compass bearing as measured for aspect.
The upper left comer was pemlanently marked to facilitate location of the plots again in the fUTUre, and all four corners were permanently tagged. Sketch maps were completed for each plot showing location of comers and individual tree ferns within, and digital photos were taken of each plot. Cibotium chamissoi were detemlined to be inside the pl ot
44 • • if the center of the growth apex was within the string-delineated plot edge at ground level. Each individual was permanently tagged with loose-fitting wire to accommodate any increase in trunk circumference, and assigned a unique identification number for future re-measure. Because it often appeared that mature individuals had multiple trunks
(caudexes), fallen leaves and other organic materials were cleared away to trace branching trunks to their origin: if multiple trunks appeared to have originated from the same rhizome, they were counted as multiple trunks of one individual; if the trunk seemed to originate separately from the soil, or if they had completely separated from the originating rhizome and were now rooted in the soil, they were tagged as separate individuals. A "trunk" was defined as having a formed caudex and presence of full-sized fronds, whereas a "trunk bud" was defined as an obviously forming lump covered with soft hairs protruding from the trunk with or without emerging crosiers and fronds. For individuals with multiple trunks, the oldest trunk with live fronds (almost always the longest trunk) was used for the measurements described below.
The following data for each C. chamissoi occurring within the research plot were collected to determine size class and morphologic variation within the sampled population (similar to methods used in Becker 1976; Durand and Goldstein 2001): tree fern caudex length (vertical and horizontal) from ground to apical meristem, trunk circumference at both just below fronds at the apex and as close to the ground as possible, height offem canopy from ground, number of trunks, number of trunk buds, length oflongest frond, total number offronds, number of unfurling crosiers
(fiddleheads), number of fertile fronds (with sori), number of healthy fronds (> 50%
45 green), and number of dying fronds « 50% green). Senesced fronds that were completely brown and retained as a skirt were not included in these measures. Size class was ranked as mature, immature, or spore ling. A sporeling was defined as having a total trunk length less than 15 em. An individual was ranked immature ifthere were no live or dead fertile fronds. "Tiny sporelings"---considered those without measurable caudexes and usually less than 5 em tall-were counted and included on plot sketch maps, but not permanently marked or measured due to their very small and fragile nature. Vigor for each measured individual C. chamissoi was ranked for as healthy, moderate, or poor. Dead individuals were not counted or measured in this study. See Appendix I for data sheets used in field data coIlection.
To address research questions two and three in determining whether certain environmental factors are correlated to C. chamissoi recruitment between sample sites, the following data were collected. Overstory (plants over 2 m tall) and understory species lists were compiled for the area based upon a visual survey, and estimated percent cover by overstory species was assigned. These estimates covered the visible area within and surrounding the plot in order to generally characterize the vegetation composition.
Similarly to generally characterize the area, pig activity was ranked for observed digging, browsing, tracks/trails, and scat. Understory species diversity-and respective percent cover was measured in five randomly located lxl m sub-plots within each plot. The sub plots were read to 1 m above the forest floor, so cover measures often exceeded 100% due to the overlapping vertical layers of vegetation. At each random understory plot location, the canopy density was measured with a densiometer in north, south, east, and
46 west directions and the average taken. Additionally, elevation, slope, aspect, and substrate were recorded for each plot, and plot locations were recorded with a GPS unit.
Two randomly located soil samples were collected with a 15 cm depth soil corer within each plot, combined, and sent to the Agricultural Diagnostic Service Center within the
University of Hawai 'i-Manoa Department of Agronomy and Soil Science for analysis of soil pH, calcium, nitrogen, organic carbon, phosphorous, potassium, and magnesium.
·To address research question number four, the vegetation physically occurring upon individual C. chamissoi trunks, including bryophytes, were noted as presence or absence of epiphytes.
Cibotium Distribution and Habitat
Research question five required a search for available spatial records of Cibotium on
O'ahu. Utilizing database records from Army Natural Resources and the Hawai'i Natural
Heritage Program of rare plants and associated species, searches were conducted to extract GPS points of Cibotium spp. observations. GPS points reflecting observations of
Cibotium spp. were saved as shape files and then plotted on a topographic map ofO'ahu with ArcMap GIS software. The rainfall data layer was obtained from Dr. Tom
Giambelluca and added to the final map.
Outplanting Trial
The following experimental outplanting of C. chamissoi was conducted to answer research question number six. Fifteen small C. chamissoi individuals were purchased
47 from Ui'au Hawai'i Nursery, where Kay Lynch propagated 13 individuals from spores and two individuals were collected as small trunk buds and grown in pots. Kahanahaiki was selected as an appropriate site for outplanting due to the easy access, existing fence, historical weed management, and presence of a healthy population of naturally occurring
C. chamissoi. Because survivorship of outplanted C. chamissoi was to be tested in an area targeted for restoration, the outplanting site was chosen in an area that had received intensive weed management. Strawberry guava (Psidium cattleianum Sabine) had been girdled and treated with herbicide in the area, and several dead but standing guava trees composed a significant portion of the overstory. Because they lacked foliage, light levels were high in this area. The native tree Pisonia spp. is also common in this area, forming a portion of the overs tory and some smaller regenerating individuals also occur in the understory and subcanopy.
Fifteen small C. chamissoi were monitored for pests and disease at the ANR baseyard nursery in Wahiawa for I month, and then were transported to the N .1.K.E. nursery facility near Kahanahiiiki and Pahole for acclimation. The ferns were flown to the planting site via helicopter sling load in a protective wind-proof box, planted, and watered on February 22, 2006.
Overstory density was measured at each planted C. chamissoi location with a densiometer in north, south, east and west cardinal directions and the average taken. One
Ixl m understory plot was also read immediately adjacent to each planted individual, oriented directly north of each fern. Individuals were measured and digital photos were
48 taken. See Appendix 2 for measured overstory density and understory density at each individual and data on trunk length, basal trunk circumference, length oflongest frond, number of healthy and dying fronds, number of crosiers, canopy height, and pot size for each individual C. chamissoi at the time of outplanting.
Statistical Analyses
Microsoft Excel was used for data entry and SPSS statistical software version 14.0 (2005) was used for all statistical analyses. Pearson product-moment correlation coefflcient was used to explore management and environmental relationships to size class, as measured by the continuous variable oftotal trunk length per individual C. chamissoi (vertical and horizontal trunk lengths summed). Spearman's Rank Order Correlation was utilized when recruitment was assessed by ranked size class (I = mature, 3 = immature,S = spore ling).
Population structure was also assessed as relative abundance by analyzing each size class as a percentage of the total C. chamissoi within each plot.
Environmental relationships were investigated using Pearson product-moment correlation coefficient across all C. chamissoi measured (n=337) and averaged per plot (n=16). Size class was ranked from mature = I, immature = 3, sporeling = 5, and abundance, relative abundance, and trunk length were also analyzed. Native and invasive overs tory and understory percent cover were calculated as follows. All overstory species present were coded as native, introduced, or invasive and estimated percent covers were totaled.
Understory species were listed, coded as native, introduced, or invasive and percent cover was totaled across all five I x 1 m understory subplots. The total cover by species was
49 divided by 5 for the average percent cover by species across the research plot, and a total
percent understory cover was then summed across all species. Total understory cover and
native understory cover were then analyzed with and without C. chamissoi data as measured in the understory sub-plots.
One-way between-groups analysis of variance (ANOVAs) were conducted to explore the effects of fencing and• weed control separately on the dependent variables of total C. chamissoi, total sporelings, total immature, and total mature individuals per plot, and relative abundance by size class per plot. Additional one-way ANOVAs were conducted to assess the effects of fencing and weed control separately on the dependent morphological variations among all C. chamissoi within the paired fenced/unfenced plots.
A two-way between groups ANOV A was also conducted to explore the combined effects of fencing and weed control on mean total trunk length. All C. chamissoi in the ten fenced/unfenced paired plots were assigned to groups according to fencing condition (1 = fenced, 2 = unfenced) and weed control condition (I = weeded, 2 = not weeded).
All 337 measured C. chamissoi individuals and total abundance in each size class of the
16 research plots were subjected to principal components analysis (PCA). Prior to performing PCA the suitability of data for factor analysis was assessed. Interpretation of the components was aided by Varimax rotation. A standard multiple regression was then conducted to explain the most variance in C. chamissoi sporeling relative abundance across all plots using the factors identified with PCA to be most variable between plots.
50 Multicollinearity of the variables was assessed with Pearson correlation and the collinearity diagnostics of Tolerance and VIF. A Nonnal Probability Plot and Scatterplot were generated to assess the linearity and nonnality of the data set. R Square values and significance were reported for the multiple regression model. Standardized coefficients and part correlation coefficients were reported for each predictor variable.
51 CHAPTER 3. REsULTS
Population Structure and Morphological Variation
Research question I: What are differences in C. chamissoi population structure and morphology between paired fenced and unfenced plots in the Wai'anae Mountains, and between all unfenced plots in both the Wai'anae and Ko'olau Mountains?
All Measured C. chamissoi
A total of337 individual C. chamissoi were measured in 16 research plots. Average morphological attributes by size class (mature, immature, sporeling) are presented in
Table 3 for all individuals measured. Of the total 337 tree ferns measured, there were 201 mature, 85 immature, and 51 spore lings.
52 Table 3.
All Plots: Average Size Class Morphology
Mature Immature Sporelinl( Mature Immature I Sporelin~ Basal Trunk Circumference (cm) Canopy Height (m) Mean 44.68 20.67 8.64 Mean 1.94 1.01 0.38 Maximum 157.50 44.00 15.00 Maximum 4.29 1.90 0.77 Minimum 21.00 11.00 2.00 Minimum 0.37 0.33 0.10 Trunk Top Circumference (cm) Longest Frond Length (m Mean 40.24 19.80 8.38 Mean 2.31 1.35 0.54 Maximum 74.00 31.00 15.00 Maximum 4.47 2.32 1.11 Minimum 24.00 0.00 0.00 Minimum 1.16 0.33 0.14 Vertical Trunk Lengthkm) Number oiHealtlf}'..Fronds Mean 53.87 15.15 4.50 Mean 7.46 3.04 2.59 Maximum 220.00 48.00 12.00 Maximum 25.00 13.00 6.00 Horizontal Trunk Lenzth (em) Minimum 0.00 0.00 1.00 Mean 60.24 13.36 2.18 Number ofFertile Fronds Maximum 400.00 84.00 11.00 Mean . 3.13 0.00 0.00 Total Trunk Lenzth (em) Maximum 13.00 Mean 112.37 27.71 6.38 Minimum 0.00 Maximum 400.00 93.00 14.00 Number ofCroziers Minimum 17.00 0.00 0.00 Mean 3.19 1.18 0.59 Number of Trunks Maximum 14.00 6.00 2.00 Mean 2.01 1.05 1.02 Minimum 0.00 0.00 0.00 Maximum 11.00 2.00 2.00
Number 0 Trunk Buds Tota/lndividuals Mean 1.73 0.26 0.04 I 201.00 85.00 I 51.00 Maximum 13.00 . 4.00 1.00
Paired FencedlUnfenced Plots
Population structure and abundance were markedly different between the five replicates of paired fenced and unfenced plots (see Figures 9 and 10), with higher numbers of total
C. chamissoi occurring in all of the fenced plots. Sporelings were present in four fenced
53 plots, but lacking in the fenced Pahole 2 site and occurring there in the unfenced plot.
Sporelings were lacking in the remaining four unfenced plots. Immature individuals occurred in all five fenced plots, and also in the unfenced plots at the Pahole I and
' Ohikilolo sites. Results of soi l analyses for paired fenced and unfenced plots are included in Appendix III.
Figure 9.
Population Structure and Abundance: Paired fenced and Unfenced Plots
"o ~~~------.
120
100 o Maru re
• Spurdmg
~ .. 1-J.cJ------'====--j ....
'0
20
54 Figure 10.
Population Structure by Percentage: Paired Fenced and Unfenced Plots
100',
o Mature ~'. 5l1mmature:
110"/. • Sporeling
70"/. 8 Tiny Sporeling ·s .,.. .~• ~ 'ij sm'. \! C •~ "-• .,."
30%
"'""
10""
""1 OJillo" ) I'OI:l'!S Kah:anahlW Piol
Results of a one-way ANOYA found the relative abundance of immature C. chamissoi to be significantly higher in fenced areas than unfenced areas (see Table 4). However, abundance and relative abundance of sporelings, mature individuals, and total C. chamissoi were not significantly different between fenced and unfenced plots.
55 Table 4.
ANOVA Results: FencedlUnfenced Immature C. chamissoi
Relative Abundance
Source 0/ Variation Sum a/Squares d/ Mean Square F P value
Between Groups 0.097 0.097 5.332 0.050
Within Groups 0.146 8 0.018
Total 0.243 9
See Figure 11 for population structure and abundance averages between all fenced and unfenced plots.
Fi gure II.
Average Population Structure and Abundance: Fenced and Unfenced Plots
50 o ~hlurc &l lmmature
45 • Sporcling
40
35 .,, ~• JO -.; "0 25 ~ E = Z. 2(l u " I;
10
0-'----- FenCM Unfenced 56 Among the paired plots, one unfenced plot has a history of weed control and was sampled at the 'Ohikilolo research site. The remaining four unfenced plots have not been subject to weed control. In addition, two fenced research plots do not have a history of weed management, at the Pahole I and Pahole 2 research sites, while the remaining three fenced plots have been subject to weed control. Correlational relationships between size class and the management variables of weed control and fencing were analyzed for all C. chamissoi found within these 10 paired research plots. Unfenced areas had tree ferns with longer trunks (r = .277, n = 169,p<.OI), and plots inside offences had a higher abundance of smaller individuals (r = -.234, n = 169, p<.O I). Areas with weed control were associated with tree ferns of shorter trunk lengths and areas without weed control were associated with tree ferns of longer trunk lengths (r = .275, n = 169, p<.OI).
Results of a one-way ANOY A indicate an average of larger basal trunk circumference, longer total trunk length, and higher average number of trunk buds in areas without previous weed control across all measured C. chamissoi in the paired fenced/unfenced plots (see Table 5).
57 Table 5.
ANOVA Results: Weed Control and Morphological Variation
Source of Variation Sum ofSquares df Mean Square F P value
Basal Trunk Circumference
Between Groups 6426.81 1 6426.81 15.872 0.000
Within Groups 65595.13 162 404.91
Total 72021.94 163
Trunk Length
Between Groups 73191.88 1 73191.88 13.635 0.000
Within Groups 896454.44 167 5367.99
Total 969646.33 168
Number of Trunk Buds
Between Groups 76.93 1 76.93 16.453 0.000
Within Groups 780.89 167 4.68
Total 857.82 168
Two-way ANOVA results in Table 6 demonstrate the similar significant effects of fencing and weed control on C. chamissoi trunk length, but no significant interaction between these two management variables (see Figure 12).
58 Table 6.
Two-way ANOVA Results: FencinglWeed Control
and C. chamissoi trunk length
Source o/Variation Type!!! d/ Mean Square F P value Partial Eta Sum 0 S uares S uared Fence 36513.99 1 36513.99 7.258 0.008 0.042
Weed 38082.46 1 38082.46 7.570 0.007 0.044
Fence*Weed 6646.82 1 6646.82 1.321 0.252 0.008
Figure 12.
Two-way ANOVA: Mean Trunk Length by Fencing and Weeding
..s:::..... 180 e.o c Q) 160 ~ ~ 140 c .-. -- Weeded ::l S 120 ~ ~'"' u -r- Not Weeded "-' / -.....~ 100 0 ~ 80 / c ~ r I Q) 60 I ~ Fenced Unfenced
59 See Table 7 for a summary of means and one-way between-groups ANaYA results of significance for C. chamissoi morphological variance by size class between fenced and unfenced plots. Mature C. chamissoi showed larger sizes in unfenced plots. However, immature and sporeling C. chamissoi generally had larger values for these morphological measures inside the fences.
60 Table 7.
Mean Differences in Morphology: Fenced and Unfenced Plots
I Fenced I Unfenced BaJa/Trunk Cimmyerrna (em) Mawre lmmarurc Sporeling Mature lmmarure Sporeling F p Mean 42.81 22.24 10.54 55.76 18.63 2.50 22.54 0.00 1\ 73 34 12 39 4 2 Std. Deviation 11.13 6.77 1.94 26.06 9.01 0.71 j/,rticoi T nlnk Lmglh (em) M 1 S M I S F P Mean 53.22 17.57 6.65 75.12 14.00 0.50 16.03 0.00 N , 74 35 13 39 3 2 Std. D eviacion 36.13 9.83 2.25 57.42 8.00 0.00 TOla/Tmnlt:. u ll}!/h (em) M I S M I S F P Mean 118.13 29.13 7.88 146.29 23.00 0.50 13.84 0.00 N 76 35 13 39 4 2 Std. D eyjation 74.02 25.00 1.73 52.1 7 23.80 0.00 COflO.~ H'/~hl (m) M 1 S M 1 S F P Mean 1.94 1.09 0.47 · 2.05 0.81 0.22 4.71 0.03 N ] 6 35 13 39 4 2 Std. Deviation 0.57 0.37 0.20 1.06 0.42 0.02 N Nmbtr of H,oi1i!J rrondr 11 1 S M I S F P ivlean 6.39 3.43 2.38 9.23 2.25 5.50 16.31 0.00 :-; 76 35 13 39 4 2 Std. De\'iation 3.99 2.42 0.77 6.85 0.50 0.7 1 j'iuHlber o/Cro~erl M I S M I S F P .Mean 3.33 1.54 0.69 4.10 1.00 0.50 5.94 0.02 N 76 35 13 39 4 2 Std. Deviation 2.37 1.22 0.63 3,63 0.82 0.71 I'.J,,,,,,ber ofTI1mle.r M I S M I S F P Mean 1.61 1.09 1.00 2.69 1.00 1.00 21.64 0.00 N 75 33 11 39 4 2 Std. D eviation 1.0767 0.291937 0 2.05393 0 0 N umber ojTnm /e Buds 1>1 I S M I S F P l\·1ean 1.64 0.31 0.00 2.72 0.75 0.00 12.11 0.00 IN 76 35 13 39 4 2 Std. Deyiaoon 2.20 0.90 0.00 2.91 0.96 0.00
6 1 All Unfenced Plots
Population structure and abundance across all unfenced plots showed major differences between the two mountain ranges (see Figures 13 and 14). Tiny sporelings, sporelings, and immature C. chamissoi were present in all Ko'olau unfenced plots, whereas tmy sporelings were present in only one ofthe seven Wai'anae unfenced plots, specifically at the Pahole 2 research site. Sporelings were present m the Pahole 2 site as well as the steep Kahanahaiki plot, but lacking in the remainmg five Wai'anae research plots.
Immature C. chamissoi were absent from three of the Wai'anae plots. Total abundance was much higher in three of the Ko'olau range plots than all of the Wai'anae plots. The
'Aiea plot (Ko'olau range) had slightly lower abundance than the 'Ohikilolo plot
(Wai'anae range) but was greater than all of the other Wai'anae plots. Statistical relationships between all unfenced C. chamissoi and environmental variables will be explored in the following sections.
62 Figure 13.
Population Structure and Abundance: All Unfenced Plots
100
<)()
~ Imm:l.!ure
80 • Sporclirlg
8 Tin)' Spore Eng 70
'5 60 . ~• •E "u ;0 <..i • ,.."$ '"
:;0
20
HI
0 B- N E !io ~ :§ • :!! ~ "• .. 2 ~ t ~ .. ~ -5 ~ ; ~ .ii" ~ ;: ",;; £ ~ .&" ...... " :< i ~ '" < " ;: 1~ 6 ;: ~
Wai'an1(' R~nRe K o'oJ~u Range. Pl ot
63 Figure 14.
Population Structure by Percentage: All Unfenced Plots
o Marorc
~ irnrTl2rure
• Sporehng
=TUl)' Spording
o
Plot
C. chamissoi Environm ental Recruitment Correlations
Research question 2: Which environmental factors are correlated to C. chamissoi recruitment between sample sites: elevation, annual rainfall, slope, aspect, soil nutrients, higher percentages of native (vs. alien) vegetation, and/or percent canopy cover?
All Measured C. chamissoi
Across all measured C. chamissoi, the following correlations between environmental variables and recruitment were observed. Slope was significantly positively correlated 64 with number of trunks per mature individual (r = .200, n = 200, p<. 0 I) and mature . number of trunk buds (r =.142, n = 20 I, p<.OS), but not related to mature horizontal trunk length. Aspect had no significant relationship with total trunk length or size class across all measured tree ferns. There was a significant negative correlation between total trunk length and rainfall (r = -.278, p<.OI) and a positive correlation between size class (ranked mature = I to sporeling = S) and rainfall (r = .174, p<.O I), meaning that there was a higher number of smaller, less mature tree ferns in areas of hi gher rainfall (see Figure
IS). However also across all C. chamissoi, elevation was negatively correlated with size class (r = -.263 , p<.01) and positively correlated with total trunk length (r = .331,p<.01).
When data were analyzed by plot average (n = 16), the above relationships were not significant, with the exception of elevation and total trunk length (r = .S 62, p<.OS).
Figure 15.
Correlation between Trunk Length and Rainfall: All C. chamissoi
4500 4000 ~---.-.. •• • • ? 3500 .§.. 3000 • 3 2500 = •• • . " 2000 = 0: -; 1500 -- ...... • • "c 1000 -;- ---- R 0.07 1< ~ 500 o o 100 200 300 400 500 Total Trunk Length (mm)
Results of soil analyses included negative correlations between individual total trunk length and soil pH (r = -.143, p<. 0 I), between size class and soil calcium (r = -.2S3, p 65 <.01), between size class and soil carbon (r = -.137, p <.05), and between size class and soil nitrogen (r = -.215, p <. 01). Positive correlations were observed between total trunk lengtb and soil calcium (r = AIO, p<.OI), between total trunk length and soil magnesium
(r = .183, p <.01), between total trunk lengtb and soil carbon (r = .247, P <.0 I), and between total trunk lengtb and soil nitrogen (r = .312,p <.0 I). Elevation also showed a strong relationship witb soil nutrients, including negative correlations witb pH (r = -
0.657, p<0.0 I), potassium (r = -.161 , p <. OI) , and magnesium (I' = -.108, p<. 05), and positive correlations with phosphorous (r = .140, p=.OI) and calcium (r = .359, p<.OI).
Soil carbon and nitrogen did not demonstrate a relationship witb elevation. When data were analyzed by plot average (n = 16), soil calcium continued to demonstrate a strong relationship witb age class, where higher relative abundance of mature individuals was associated witb higher levels of soil calcium (r = .582, p<.05), but tbe opposite relationship was observed between soi l calcium and relative abundance of sporelings (r =
-.536, p<.05) and abundance of immature individuals (r = -.508, p<. 05). Similarly, nitrogen was negatively correlated to relative abundance of immature individuals (r = -
.513, p<.05).
Total understory percent cover was found to negatively correlate witb size class both including C. chamissoi (r = -.163, p<.0 I) and excluding C. chamissoi (r = -.165, p<.O I) from the analysis. In other words, areas with numerous mature individuals had higher total understory cover (see Figure 16). Similarly, native understory cover was higher in areas witb more mature individuals as analyzed both with C. chamissoi (r = -.154, p<.O I) and without C. chamissoi (r= -.148,p<.01).
66 Figure 16.
Correlation between Understory Cover (Excluding C. chamissoil
and Size Class
180 ~ ~ • • Q '.., ~ c 160 ~'" w .~ • • .,.., 0 120 I U -<::'"' I ! • >. <.> 100 • • o (.j • • R2 _ 0.0271 ~ 80 '"w bl) • c: • "0'" 60 • • • c: "0 ;:l 40 U" !S >< 0 ~ 20 ~ l-< 0 o 1 2 3 4 5 6 Size Class: l=M ature 3=lmmature 5=Sporeling ------
Longer trunk lengths were correlated both with higher native understory cover when the
C. chamissoi understory cover was excluded from analysis (r = . 127, p<.05) and a lower invasive understory cover (r = -.107, p<.05) (see Figure 17). Average overs tory density was higher in areas with longer C. chamissoi trunk lengths (r = .111, p<.05), and across plot averages (n = 16) spore lings were more abundant in plots with hi gher native overstory cover (r = .575, p<.05). There were no other significant correlations observed across plot averages for abundance/relative abundance by size class or total trunk length.
67 Figure 17.
Correlation between Invasive Understory Cover and Trunk Length
140 " > "c . 0 u 120 ~. .-- ~ E 100 0 0 0 E 0 000 0 80 0 0 0 -0" - ••••00 0 0 C ~ - ;:>- - 60 0 o 0 _00• • 0 0 "> ",.; 40 00 0 •> ~~-. . R - 1JlJ1Tj" .: . 0 0 0 ... 20 ~:' ..":' j""~ <5 .. _._ 0 f- 0 o 100 200 300 400 500 Total Trunk Length (mm)
All Unfenced Plots
Individual C. chamissoi, as measured in all unfenced plots, were analyzed and significant correlati ons were observed between the variables as presented in Table 8.
68 Table 8.
Correlates: A ll Unfenced C. chamissoi
Rainfall Slope Aspect pH Ca Mg N Pig Native Overs/ory Activity Overs/my Del1sity Cover Size Class r = .184 r = . 161 r = .199 r = -.21 5 r = .163 r = .152 11 = 213 11 = 213 11 = 213 n = 213 n = 213 11 = 2 13 p <.01 p <.05 p<.OI p<.OI p<.05 p<.05 Trunk Length r = -.35 1 r = .164 r = .416 r = .162 r = .177 11 = 213 n = 213 11 = 213 n = 2 13 n = 213 p<.OI p<.05 p<.OI p<.05 p=.OI Number of r = -.36 1 r = .304 Trunks 11 = 213 11 = 213 p<.OI p<.OI Basal r = .1 7 1 Trunk 11 = 209 Circumference p<. 05 Vertical r = -. 139 r = .146 Trunk 11 = 211 11 = 2 11 Length p<.05 p<.05 ,
69 Size class and rainfall were positively correlated (r= .184, 11 = 2l3, p<.01) as were size class and slope(r = . 161 ,11 = 213, p<.05). Total trunk length was positively correlated with aspect (r = .164, 11 = 213, p <. 05), meaning south and west facing slopes had larger tree ferns, with aspect ranked from I = NNE, 2 = ENE, 3 = ESE, 4 = SSE, 5 = SSW, 6 =
WSW, 7 = WNW, 8 = NNW. Total trunk length was also negatively correlated with rainfall (r= -.351 , 11 = 213, p<.01). Slope was not related to horizontal trunk length, but was negatively correlated with vertical trunk length (r = -.139, 11 = 211, p<. 05). Number of trunks were positively correlated with ranked aspect (r = .304, 11 = 213, p <. OI ), meaning south and west facing slopes had tree ferns with an increased number of trunks.
Rainfall was negatively correlated with number of trunks (r = -.361, n = 213, p<.O I).
Results of soil analyses for all unfenced plots are included in Appendix IV. Soil analyses indicated positive correlations between size class and pH (1' = .199, 11 = 213 , p<.01) and total trunk length and soil calcium (r = .416, n = 213,p<. 01). Total trunk length was also positively correlated with soil magnesium (I' = .162, n = 213, p<.05) and soil nitrogen (r
= .177, n = 213, p=.O I) Size class was negatively correlated with soil calcium (r = -.215, n = 213,p<.01 ). Pig activity was positively correlated with basal trunk circumference (r=
.171 , n = 209, p<.05) and vertical trunk length (r = .146, n = 21 I, p <.05), but negatively correlated with number of trunks (r = -.215, n = 213, p<.OI ). Cover estimates were related to recruitment across all unfenced C. chamissoi as follows. Size class was positively correlated with native overstory cover (r = .163, 11 = 213, p<.05) and total overs tory density (I' = .152, n = 2 13, p <.05). 0 significant relationships were found between recruitment and total understory cover or native/invasive understory cover.
70 All II unfenced plots were separately analyzed across plot average (n = II in all cases) and significant correlations were observed between the variables as presented in Table 9.
Aspect, pig activity, soil nutrients, canopy density, and native/invasive overstory cover were not significantly correlated with recruitment, nor were total number of invasive species. Total understory cover was positively correlated with total tiny sporelings (r =
.751 , p<.01), however not when C. chamissoi were removed from analysis, and similarly total immature were positively correlated with native understory cover (r = .690, p<.05), but not when C. chamissoi were removed from cover estimates.
Ta ble 9.
Correlates: All Unfenced Plots*
Elevation Rainfall Slope Total Total Native UnderstOlY Unders!OIY Species Species Total 1'=-.611 p <.05 Sporelings
Percent r = -.677 p<.022 Sporelings
Total r = -.779 l' =-.602 r =-.639 p <.01 p=.05 p <.05 immature Percent r= .702 r =-.685 p <.05 p <. 05 Mature Trunk Length r = .834 r =-.621 r= .690 r = .718 P <. OI p <.05 p <.05 p <.05 * n = 11 in all cases
71 I
Variance in C. chamissoi Recruitment and Abundance
Research question 3: Using principal components analysis (PCA), which measured variables are most responsible for variance in collected data among research plots: slope, aspect, rainfall, fencing, weeding, native or invasive percent understory cover, native or invasive percent overstory cover, overstory density, soil nutrients, or pig activity? Using multiple regression, which of these variables explain the most variance in C. chamissoi sporeling relative abundance across all plots?
An inspection of the PCA correlation matrix revealed the strongest relationships between the following 10 variables: fencing, weeding, average overstory density, native overstory cover, understory cover excluding C. chamissoi, average native understory cover excluding C. chamissoi, slope, annual rainfall, pig activity, and percentage of sporelings.
The correlation matrix of these 10 variables consisted of many coefficients of.3 and above. The Kaiser-Meyer-Oklin (KMO) value was found to be .546, which achieved the recommended value of.6 for sampling adequacy (Kaiser 1974). The Bartlett's Test of
Sphericity reached statistical significance (Bartlett 1954). Multiple trials of additional variables including soil nutrients, aspect, elevation, invasive overstory cover, invasive understory cover, trunk length, total C. chamissoi, total sporelings, total immature, total mature, percent immature, and percent mature resulted in a reduced KMO value.
Principal components analysis revealed three components with eigenvalues exceeding 1, explaining variance respectively of39.5 %, 18.7%, and 16.4%, for a total explained variance of74.6% (see Table 10).
72 Table 10.
Principal Components Analysis Total Variance
Initial Eigenvalues Extraction Sums of Squared Loadings Component Total % of Variance Cumulative % Total % of Variance Cumulative \1/0 1 3.947 39.473 39.473 3.947 39.473 39.473 2 1.871 18.711 58.183 1.871 18.711 58.183 3 1.636 16.362 74.546 1.636 16.362 74546 4 0.741 7.413 81.958 5 0.627 6.271 88.229 6 0.412 4.119 92.348 7 0.330 3.304 95.652 8 0.258 2.581 98.233 9 0.132 1.317 99.550 10 0.045 0.450 100.000 Extraction Method: Principal Component Analysis.
eatell's (1966) scree test was utilized to support retaining three components for analysis, and the screeplot reveals a clear break between the third and fourth components (see
Figure 18). Results of Parallel Analysis further supported the retention of three components, which showed these components having eigenvalues exceeding the corresponding criterion values for a randomly generated data matrix of 10 variables x 337 individuals.
73 Figure 18.
Principal Components Analysis Scree Plot
Scree Plot
, , Component Number
The Varimax rotation of the three components revealed unique variables strongly loading on each component. However, three variables loaded on multiple components. These included fencing and native understory cover across all three components, and the variable of average overs tory density loaded on two components. The three-component solution explained 74.6% of the data variance: component 1 contributed 29.7%, component 2 contributed 25.8%, and component 3 contributed 19.1 % (see table 11).
74 Table 11.
Varimax Rotation of Three Factor Solution for Variance:
Pattern/Structure for Coefficients
Component Variable 1 2 3 Slope 0.815 Average Overstory Density 0.807 -0.368 Weeded 0.755 Understoty Cover (ex Cibotium) -0.685 Fence 0.642 0.469 -0.330 Pig Activitv 0.863 Annual Rainfall 0.825 Native Understory Cover (ex Ciborium) -0.417 -0.701 0.447 Percent Sporelings 0.856 Native Overstorv Cover 0.837
Total % Variance Explained 1 29.71 25.81 19.1
The greatest variation in the data among plots was explained component 1 with the variables of slope and overstory density as the highest loading variables. Pig activity and annual rainfall were highest loading on component 2, and percent sporelings and native overstorycover were highest loading on component 3. These variables were selected for exploring further with a standard multiple regression analysis. See Table 12 for the correlation matrix between the selected variables, noting they are generally poorly correlated. Together the variables of slope, overstory density, pig activity, annual rainfall, and native overstory cover explain 34.9 percent of the variance inC. chamissoi sporeling relative abundance across all plots and the multiple regression model achieved statistical significance (p<.Ol) (see Table 13). Total native overstory cover contributed the most to the prediction of sporeling relative abundance, and uniquely explained 25.4 percent of the
75 variance in sporeling relative abundance as detennined by squaring the part correlation coefficient (see Table 14).
Table 12.
Pearson Correlates: Multiple Regression Model Variables
Percent Slope Overstory Pig Rainfall Native Sporelings Density Activity Overs tory Cover Percent 1.000 -0.136 0.032 0.140 0.202 0.506 Sporelings Slope -0.136 1.000 0.479 0.132 0.189 -0.264 Overstory 0.032 0.479 1.000 -0.238 -0.074 0.180 Density Pig Activity 0.140 0.132 -0.238 1.000 0.575 -0.160 Rain 0.202 0.189 -0.074 0.575 1.000 -0.202 Native 0.506 -0.264 0.180 -0.160 -0.202 1.000 Overstory Cover
Table 13.
Multiple Regression Results
R Square AdjustedR Std. Square Error Model Summary .358 .349 .095
Regression Sum of df Mean F Significance Sauares Square Model p<.OI ANOVA 1.667 5 .333 36.951
76 Table 14.
Multiple Regression Model Predictor Variables
Unstandardized Standardized Coefficients Coefficients B Std. Beta t Sig. Part Error Correlations Slope -0.001 0.001 -0.044 -0.773 00440 -0.034 Overstory 0.000 0.001 -0.011 -0.193 0.847 -0.008 Density Pig Act 0.007 0.006 0.071 1.262 0.208 0.056 Rain 0.000 0.000 0.282 5.155 0.000 0.227 TNOC 0.002 0.000 0.565 110457 0.000 0.504
Presence of Epiphytes
Research question 4: What is the frequency of epiphytic colonization on Cibotium chamissoi individuals and which vegetation community characteristics, environmental characteristics, and morphological characteristics predict presence or absence of epiphytes?
Among all 337 C. chamissoi measured, the presence of epiphytes (ranked I = yes, 2 = no) exhibited the strongest relationship with larger and older tree ferns. Basal trunk circumference (r = -0421, n = 328,p<.01) and trunk length (r = -0402, n = 337,p<.01) were most significantly correlated with epiphytes (see Figures 19 and 20). Environmental site characteristics of slope, aspect, and rainfall did not demonstrate a significant relationship with epiphytic presence on C. chamissoi. Higher levels of the soil nutrients carbon (r = -.210, n = 337,p<.01) and nitrogen (r = -.184, n = 337, p<.Ol) were significantly related to epiphytic presence upon C. chamissoi. Higher native overstory 77 cover indicated a less likely epiphytic colonization (r = .176, n = 337,p<.01), whereas higher non-native overstory cover was related to more frequent epiphytic presence (r = -
.211, n = 337,p<.OI). A higher average overstory density was related to an absence of epiphytes (r = .160, n = 337, p<.OJ), whereas higher understory cover (r = -.130, n = 337, p<.05) and invasive understory cover (r = -.263, n = 337, p<.OJ) were related to more frequent occurrence of epiphytes on C. chamissoi.
Figure 19.
Correlation between Presence of Epiphytes and Basal Trunk Circumference
E 180 .;:. 160 ~ u 140 ~ ~ 120 • J:l" • E 100 u 80 • • " ...iJ 60 • ~ 40 " 1-< 20 R 0.1776 .. 0 • "" 0 1 2 3 Epiphytes: 1 = yes 2 = no
78 Figure 20.
Correlation between Presence of Epiphytes and Trunk Length
450 a 400 .!:- 350 • -5 300 • • OJ) i "~ 250 ~ 200 - : I "", 150 r-"~ -; 100 0; 50 r- -R' - 0.1616 0 , a 1 2 3 Epiphytes: 1 =yes 2 = no
Cibotium Distribution Observations
Research question 5: Where has Cibotium been recorded to occur on O'ahu by the Army
Natural Resources and Hawai'i Natural Heritage Program?
See Figure 21 for results of GPS records of Cibotium observations from Army Natural
Resources and the Hawai'i Natural Heritage Program. Observations often coincided with rare plant populations, therefore this mapping exercise should not be considered representative for total range of Cibotium on O'ahu, nor can most appropriate habitat for
Cibotium be predicted from this data. It should be noted that the 16 research plot locations of this study were also included in Figure 21, and could be considered more representative of the much greater range of Cibotium. Ranges of Cibotium observations occur between 1,500 to 7,000 mm ammal rainfall. Additional data for Cibotium
79 collections with Bishop Herbarium have been obtained, but because verbal directions to locations of collections were not accompanied by GPS locations, points will need to be hand plotted on topographic maps in the future and were not included in this data set.
Figure 21.
Observations of Cibotium on O'ahu
O'ahu, Hawai'i Legend
• Citbotium Observatlons o Research Plots (16) -- Annual Rainfall (mm) Elevation (m) High: 1233
Low:O
1:250,000 o 5 10
80 Outplanting Trial
Research question 6: What is the survivorship of 15 nursery grown and experimentally outplanted C. chamissoi into Kahanahaiki, a fenced area of forest restoration habitat in the Wai'anae Mountains where other C. chamissoi naturally occur?
Only one C. chamissoi has survived the outplanting trial as of March 21, 2007. This individual (plant number 392) was planted in an area of 49.8 % overstory density and the adjacent understory lxl m sub-plot contained a 25 % total cover composed by Pisonia umbellifera (20%) and bryophytes (5%). After original outplanting on February 22, 2006, all individuals were reported to be alive through the rainy winter, but high rates of mortality occurred during the dry summer months. Outplanted C. chamissoi were reported to be showing extreme signs of decline in September 2006, and by date of first re-measure on November 19, 2006 only no. 392 was confirmed to be living. Plant number 392 was one of two "trunk button" propagules, originally collected from
Kahanahiiiki 4/29/2005 and until the time of outplanting was in a two gallon-sized pot.
After 13 months, fern no. 392 grew approximately .5 em in trunk height and the longest frond length increased by 8 em. The only notable distinction for number 392 was the basal trunk circumference-at the time of outplanting it was the largest of all 15 outplanted individuals.
81 CHAPTER 4. DISCUSSION
The following chapter includes discussion on the results of the six main research questions and concludes with suggestions for future research in forest restoration using C. chamissoi. The purpose of this biogeographical study was to examine the population structure, abundance, and potential restoration importance of the endemic tree fern C. chamissoi across a spectrum of intact and degraded forest communities in the Wai'anae and Ko'olau Mountains ofO'ahu.
Effects of Ungulate Predation
Plot location was biased to include known existing groves of C. chamissoi. The goal was to determine whether the species is able to maintain itself through successful recruitment where it occurs, or whether the species may be in decline on O'ahu. In the Wai'anae
Mountains, where plots were paired inside and outside of fenced exclosures for similar elevation, slope, aspect, and vegetation associates, the impacts of feral pig predation were clearly observed. Unfenced C. chamissoi abundance was lower, and a major percentage of the unfenced population consisted of mature individuals with fewer sporeling and immature cohorts. Relative abundance of unfenced immature C. chamissoi was found to be significantly lower, which indicates survivorship of unfenced sporelings and immature individuals has been limited, and these unfenced populations may continue to decline in abundance. The reduced sizes of unfenced sporeling and immature tree ferns, and larger size of mature unfenced individuals further demonstrates limited recruitment in unfenced areas. Table 7 clearly shows significant differences in sizes of mature, immature, and 82 sporeling individuals between paired fenced/unfenced plots, with average sizes of unfenced spore1ings and immature individuals much smaller and average sizes of unfenced mature C. chamissoi greater than those within fences. Longer trunk lengths of
C. chamissoi were significantly correlated with unfenced areas, and presence of fencing was significantly related to abundance of sporelings and immature individuals. Based upon the results of the paired plot sample of this study, it would be reasonable to conclude that feral pigs are limiting C. chamissoi recruitment outside of fenced areas in the Wai'anae Mountains.
It is worth noting that a significance difference was not found for a lower abundance of unfenced sporelings. Although C. chamissoi are slow growing and large individuals are very old, the mature individuals produce several million spores per fertile frond. As measured in this study, mature C. chamissoi had an average of 3.13 fertile fronds.
Gametophyte germination and fertilization of C. glaucum have been shown to depend on shade and moisture conditions (Friend 1974), and sporeling C. chamissoi in nursery conditions thus far have demonstrated a high mortality rate, with survivorship between 20 to 30 percent (Kay Lynch, unpublished data). The first limiting factor in sporeling survivorship could therefore be considered the appropriateness of microsite habitat.
Figures 9 and 10 show that spore lings were less abundant in the fenced Kahanahaiki and
Pahole 1 plots, and the fenced Pahole 2 plot did not contain any sporelings at all. The
Pahole 2 site was interesting in that the Ulifenced plot did contain both tiny spore lings and spore lings, but no immature C. chamissoi. It does not seem that fence age (or years of protection from ungulate predation) explains variation in sporeling abundance or C.
83 chamissoi abundance in general. Because the fenced exclosures at the Kahanahiiiki and
Pahole sites were pig free in 1998, C. chamissoi had been protected from pig predation there for the longest of the measured fenced populations-approximately 8 years at the time of this study. The 'Ohikilolo fence was constructed one year later in 1999, and this site had the highest abundance offenced C. chamissoi and the highest abundance of fenced sporelings. The 'Ohikilolo site was most likely free from feral pigs even prior to the fence construction due to the steep cliff terrain surrounding the site. The previous browsing pressure of feral goats at 'Ohikilolo may have limited survivorship of sporeling
C. chamissoi prior to fence construction, but would not have caused the higher rates of mortality for immature and mature individuals as observed with feral pig predation. The
Three Points fence was constructed and the area made pig free most recently in 2001, but this plot had the second highest abundance of C. chamissoi in the Wai'anae Mountains and the largest immature cohort of all fenced plots. Had the temporal dynamics of C. chamissoi recruitment coincided with fence age as expected, the fenced Pahole and
Kahanahiiiki plots would have contained a bumper crop of sporeling and immature individuals. Thus it seems protection from ungulates is related to C. chamissoi abundance and recruitment when compared between paired fenced/unfenced plots, but recruitment and survival of sporelings is also influenced by other factors across the research sites.
Effects of Weed Control
Previous weed control across the paired fenced/unfenced plots is difficult to assess independent of the effects of fencing and, though not the main focus of this study, was an important variable to include because it has possible implications for land management.
84 Only one of five nnfenced plots had a history of weed control and two offive fenced plots had never been weeded, therefore the data from this study cannot make any definite conclusions regarding weed control effects on C. chamissoi recruitment. However, limited results indicate removal of invasive vegetation may increase recruitment, as a significant correlation was found between weed control and abundance of smaller individuals, and significance was found in a one-way ANOVA for areas without weed control to contain larger individuals. These results were likely compounded by the effects offencing, but are worth exploring further in future research. Figure 12 illustrates two way ANOVA results for the independent variables of fencing and weeding, and these results show the effects of each separately on average trunk length. The shorter average trunk length in areas of weed control indicates higher rates of recent recruitment, with larger cohorts of spore ling and immature C. chamissoi causing the average trunk length to be shorter than areas with predominantly larger and older individuals. Figure 12 shows the significant effects of weed control on recruitment both in fenced and nnfenced C. chamissoi populations, and the significantly larger average size of individuals in the un weeded fenced plots. These limited results'indicate that initial weed control of areas with invasive vegetation may be necessary around both fenced and unfenced populations for management of C. chamissoi to encourage recruitment and maintain a healthy population structure.
Effects of Environmental Conditions
Alii! unfenced research plots were compared using Pearson's correlation in order to determine whether slope, aspect, annual rainfall, elevation, soil nutrients, canopy density,
85 or percent native vs. alien vegetation cover may be related to recruitment of C. chamisso(
Unfenced research sites were potentially exposed to various levels of feral ungulate browsing pressures, depending on the steepness of terrain in the case of pigs, and depending on spatially variable ungulate control efforts. Unfenced research sites may have been targeted for monitored pig control with increased hunting or snaring efforts, or may have been frequented by public pig hunters through unofficial access. Noting these potential variations in pig populations across the research sites, nonetheless significant relationships were found between ranked levels of pig activity and larger trunk circumference, longer vertical trunk length, and reduced numbers oftrunks across all unfenced plots. These correlations further support the previous conclusion that feral pigs are limiting recruitment in unfenced areas with high pig activity.
It is known that feral pigs do not frequent very steep areas, and therefore unfenced plots with steeper slopes may have been more protected from pig disturbance. Both variables of slope and pig activity were included in the multiple regression model, but their respective contribution to the variation in relative sporeling abundance was rather low.
Yet a significant positive correlation was found between all unfenced smaller individuals and steeper slopes. For example, the' Aiea Ridge and Kahana Valley plots were similar in slope, and had the least steep terrain of the four unfenced Ko'olau plots. These 'Aiea and
Kahana plots contained the lowest number of C. chamissoi among the KO'olau sites. In addition, the very steep unfenced plot at Kahanahahiki was located on a waterfall, along a rocky slope leading up to the edge of a wall of rock. This steep unfenced plot had the largest immature cohort of all Wai'anae unfenced plots, and was one of two unfenced
86 Wai'anae plots to also contain sporelings. However, C. chamissoi on steeper slopes may also be more likely to produce multiple trunks that fall away from the parent earlier in steep terrain, creating a larger immature cohort in these areas.
A significant relationship was found between areas of higher annual rainfall and higher abundance of both smaller individuals and all measured C. chamissoi across all unfenced plots. This could indicate that recruitment is influenced by annual rainfall, and this observation is further substantiated by the multiple regression analysis where annual rainfall contributed significantly and in second most importance to the variation in relative abundance of sporelings. The potential effects of global climate change on annual rainfall quantities and spatial patterns may have significant impacts to C. chamissoi abundance and distribution in the future, and may have also done so in the past.
Aspect was found to be positively correlated with total trunk length and number of trunks, though it is difficult to interpret these results. Sample sites in south or southwest facing locations were few: the majority of plots were located with north or northeast aspect. It would be useful to survey island distribution in order to determine whether C. chamissoi has a preference for certain aspects, or whether the north facing aspects sampled in this study were simply due to selective sampling methods. Without additional data, the possible habitat and morphological variation of C. chamissoi by aspect remains uncertain.
87 The relationship between larger C. chamissoi and soil nutrients revealed higher levels of soil calcium, nitrogen, and carbon were associated with areas of predominantly larger individuals. It is possible that mature C. chamissoi are either limited to these areas of higher soil nutrient concentrations or mature C. charnissoi may be having an effect on soil chemistry. Whether the negative association of smaller sized individuals in these areas was a function of a lesser ability for small C. chamissoi to tolerate higher levels of these soil nutrients, or whether the small C. chamissoi were simply less abundant due to other recruitment factors (such as a normally smaller cohort in natural population structure) cannot be distinguished.
Overstory density and native overstory cover were significantly correlated with abundance of sporelings and immature individuals for all unfenced C. chamissoi, though a similar relationship was not found for understory density or native cover. On the other hand, when assessed across all C. chamissoi populations (both fenced and unfenced) the higher understory cover and native understory cover values associated with mature individuals---despite their being removed from the cover data as measured-may indicate large C. chamissoi playa structural role in maintaining a healthy understory. However, it remains uncertain whether other factors may be responsible for epcouraging the growth of all native species, or whether mature C. chamissoi may be limited in distribution to these more native and densely vegetated areas. The negative association for sporelings and immature individuals in these areas again may have been due to the naturally smaller cohort of these size classes in regular population structure, or may be a factor of understory vegetation restraints on recruitment. These results may also indicate that
88 overstory density and native overstory vegetation are more important to recruitment than understory dynamics, as native overstory cover was the most predictive of relative sporeling abundance in the multiple regression analysis. However, assessments of all measured C. chamissoi (fenced and unfenced) for potential environmental influences on recruitment are likely compounded by the effects of fencing and weed control in this study, and thus should be interpreted with caution.
Principal components analysis results enabled identification of variables which were responsible for the majority of variation within the dataset across all research plots. These variables included slope and overstory density as the highest loading variables, along with pig activity, annual rainfall, percent sporelings, and native overstory cover. The dynamics of overstory and understory density apparently have a complex relationship with C. chamissoi recruitment, as overstory density, native overstory cover, understory density, and native understory cover were also highly variable across the plots. Together these variables grouped into three components and predicted approximately 75% ofthe variation among all research plots. Due to the nature of sampling in this study, where 16 research plots were measured and environmental data was included for each plot, PCA is limited in that environmental data was the same for each of the plots. PCA would have been a more useful and powerful analytical tool if transect data was gathered with a more continuous variation in environmental data for each C. chamissoi measured.
89 Morphological Variation
Measurements of all 337 individuals found 82 C. chamissoi (the majority of which were mature) to have more than one trunk. Thirty-eight percent of the total measured mature C. chamissoi had multiple trunks. It seems trunk buds were more common among sampled
C. chamissoi, as they were present in 40 percent of the measured population, with the majority occurring on mature tree ferns. Sixty-eight percent of mature tree ferns had at least one trunk bud. These multiple trunks and trunk buds may have been clones of the parent, or sexually germinated sporelings (from gametophytes) which then grew into epiphytic trunks. It is likely that multiple trunks and trunk buds are vegetative clones of the parent, and may comprise a significant proportion of the clumped C. chamissoi populations on O'ahu.
The horizontal trunk growth of many C. chamissoi was not linked to slope or pig predation in this study. The average horizontal trunk length of mature individuals was longer than that of vertical trunk length. The lack of correlation between horizontal trunk length and pig activity is likely due to the fatal browsing pressure of feral pigs. Once the tree fern is knocked over, pigs usually will chew through the trunk to the starchy core along the entire length of the trunk and the tree fern will die. The lack of correlation between horizontal trunk length and slope was counterintuitive because, as C. chamissoi grow taller and their fronds lengthen, the crown grows heavier. On steep slopes it could be assumed that hapu 'u would be more prone to toppling down slope, followed by re rooting and new vertical trunk growth. Indeed, the negative correlation found between vertical trunk length and slope for all unfenced C. chamissoi supports this assumption.
90 However, it seems horizontal trunk growth in tree ferns may be related to other factors
that were not examined in this study. Possible explanations for the predominance of
horizontal trunk growth could include historical events of strong winds which caused
toppling, or growth habit where a second trunk forms on an existing tree fern and at a
certain size or weight falls away from the parent and roots independently. These
hypotheses could be tested with DNA fingerprinting techniques within hapu 'u groves to
determine whether individuals are clonally reproduced, along with an examination of
historical events of high winds or hurricanes and their concurrence with tree fern age
estimates for horizontal trunk lengths .
•. Mature C. chamissoi were defined in this study as those observed with live or dead fertile
fronds. Possible age estimates for mature individuals based upon Durand and Goldstein
(2001) growth rates for C. chamissoi on O'ahu at three centimeters per year would
indicate the average age of mature C. chamissoi on O'ahu are around 37 years old, with
the largest measured individual around 133 years old. Using the same estimated growth
rate by Durand and Goldstein, average age of immature individuals would be around 9
years. An incidental determination of the transitional size of C. chamissoi from immature
to mature based upon morphological data for the 337 individuals measured in this study
can also be reported here. In situ assessment of Cibotium size at maturity has not
previously been published. The average basal trunk circumference of immature
individuals was measured at 20.67 cm, with a maximum of 44 cm. Mature individuals
were found to have an average of 44.68 cm basal trunk circumference, with a minimum
circumference of21cm, indicating the transition from immature to mature occurs
91 between 21 and 44 cm basal circumference. The total trunk length average for immature individuals was measured at 27.71 cm, or approximately 9 years old, with a maximum trunk length measured at 93 cm (31 years old). The mature individuals were measured with a minimum trunk length of 17 cm, which is estimated at 5.7 years old. This morphological data and age estimation based upon trunk length suggests that maturity occurs for C. chamissoi anywhere between 6 to 31 years of age. It seems age and size at maturity are rather variable, though age estimates of 9 years for immature individuals and
37 years for mature individuals indicate C. chamissoi is very slow-growing, late to mature, and long-lived. These characteristics make C. chamissoi more vulnerable to high rates of predation by feral pigs and other impacts such as harvesting for landscaping or orchid growing purposes. Recovery after disturbance will take many years for the slow growing C. chamissoi, and the species may never recover in areas heavily invaded by pigs and invasive vegetation whereit is continuously displaced.
Epiphytes
The results of this study indicate that larger, older C. chamissoi trunks were significantly more often colonized by epiphytes. These findings support the hypothesis by Medeiros et al. (1993) that tree fern age explains higher epiphytic occurrence on Cibotium trunks as compared to the invasive tree fern Sphaeropteris cooperi, which are younger and much faster growing. The retention of a "skirt" of dead fronds has been proposed as an adaptive strategy by Page and Brownsey (1986) for other species of tree ferns. Among Hawaiian species of Cibotium, only C. chamissoi has this habit and it may be an adaptation to protect younger, smaller C. chamissoi from becoming overgrown with epiphytes.
92 Understory density-both total density and invasive cover-were also predictors of more frequent presence of epiphytes. C. chamissoi was more likely to support epiphytes when overstory invasive percent cover was high, but less likely to have epiphytes in areas of high overstory density and high native overstory cover. These results indicate that epiphytes are colonizing C. chamissoi trunks in weedy areas. Because measurements did not differentiate native and non-native epiphytes, it is not known whether invasive vegetation is spreading to C. chamissoi trunks in weedy areas or whether native vegetation is more likely to grow epiphytically on C. chamissoi trunks for the advantage of escaping competition at the invaded forest floor.
Cibotium Distribution
The data obtained for observations and GPS points of Cibotium spp. from Army Natural
Resources and the Hawai'i Natural Heritage Program and presented in Figure 21. The purpose of obtaining these records was to map Cibotium distribution on O'ahu. However, as the data focused on rare plant populations on O'ahu that are concentrated at higher elevations, the mapping exercise included in this study cannot be considered a complete distribution. Indeed, the lower elevationallocations of the sample plots across the
Ko'olau Mountains reflects the much larger range of Cibotium than is represented by the
GPS points obtained from Army Natural Resources and Hawai'i Natural Heritage
Program. A much more comprehensive distribution map could be generated by adding records from Bishop Herbarium. However, due to the nature of the collection notes, precise locations may be difficult to identify and herbarium collection data is also subject to major biases. Aerial surveys in cooperation with Carnegie Airborne Observatory for
93 hypespectral imaging in 3-D may prove to be a useful and reliable method to map
Cibotium_spp. distribution. If such data were to be collected across the Hawaiian Islands,
comparisons could be made for elevational and topographic variations in distribution by
island. •
Outplanting Trial
In considering the suggestions by Mueller-Dombois (2005; in press) for use of C. chamissoi in restoration on O'ahu, the high mortality rates observed in the outplanting trial of small spore-raised C. chamissoi indicate that, although sporelings can be planted directly into the soil, site conditions must be carefully considered before attempting to re introduce tree ferns to an area. Shade--or understory/overstory density-seems to be an important factor that may influence the survival rates of small individuals. Because shade is directly related to micro-site atmospheric humidity and soil moisture content, openness of the understory is an important factor for consideration in using tree ferns in restoration sites. The individuals that did not survive the hot dry summer months were reported to appear dried out. Often weed control of large, dense stands of strawberry guava (Psidium cattleianum} reduce the understory and overstory cover to a significant degree, such that targeting these areas for restoration may need to involve supplemental plantings of other understory species and sub-canopy mediator vegetation to increase the shade and moisture of the site for appropriate C. chamissoi conditions. The outplanting site selected for this study was an area that had been invaded by strawberry guava and subsequently weeded to greatly reduce the overstory cover at the site. However, the overstory density and total understory cover, as measured for each ofthe 15 sporeling C. chamissoi planted
94 within the area, were not higher for the one surviving individual. The only remarkable characteristic of the lone survivor was that it was measured with the largest basal trunk circumference, and it was propagated from a vegetative trunk bud rather than from a gametophyte-germinated sporeling. It is possible that the more developed caudex and rhizome were more suited for outplanting, and vegetative propagation of trunk buds should be explored in the future as a possible way to increase stock for restoration purposes.
Future Research
Despite the useful studies of Cibotium growth rates in Hawai'i, due to the past exclusive sampling of mature individuals (caudex > I m in length), there is still a lack of knowledge about growth rates for sporeling and immature Cibotium. The assumption that growth rates remain relatively constant throughout development is problematic in that light levels at the forest floor are reduced for smaller individuals, and there may be potential differences in fern resource allocation for rhizome development at different size classes. The C. chamissoi ranked as sporeling and immature in this study will be re measured and growth rates will be determined as a part of a future research project to determine whether growth rates differ across size classes. Although growth rates may vary for the same species across islands of differing ages, there may be general trends in developmental growth rates within species or perhaps the entire genera of Cibotium. Such research will inform more accurate age estimates for C. chamissoi and will indicate whether similar studies for additional tree fern genera should be conducted.
.t 95 A future study of naturally occurring tiny Cibotium sporeling survivorship in situ is also needed, with specific measures of natural rates of sporeling recruitment and mortality conducted over a longer timeframe. It would be useful to determine whether sporeling germination and survival is related to understory or overstory gap dynamics, weed control, or natural microsite conditions such as humidity, drainage, or soil moisture. It is also important to determine the in situ survivorship rates of spore lings to immature individuals, as sporeling C. chamissoi have certainly demonstrated high mortality thus far in horticultural growth trials during this first stage of their life cycle.
Because this study targeted areas of known C. chamissoi groves, additional research of comprehensive island distribution is needed. Casual observations of C. chamissoi in this study found individuals to be clumped in distribution and were often found in drainages.
However, groves were also observed on gradual mid-slopes as well. Observations by
Palmer (1994; 2003) were found true in that C. chamissoi was the dominant species of
Cibotiumat low to mid-elevations on O'ahu. Only one small C. menziesii was observed in the research site at Kahuku, though not occurring within the plot. Stratified random transects across elevation and topographic gradients, which record presence/absence of
Cibotium and related environmental data, would be useful to determine the exact ranges and overlaps in ranges of C. chamissoi, C. menziesii, and C. glaucum on O'ahu. It would be very interesting to compare the results of distribution on O'ahu to the biogeographic patterns ofCibotium on Hawai'i, Kaua'i, and Maui. Rainfall data across the measured transects would be useful for five years previous to the study in order to determine how and where variation in annual rainfall may be affecting Cibotium recruitment.
96 Currently there is no knowledge on the predominant means of natural in situ reproduction for C. chamissoi, i.e. whether vegetative or sexual strategies are most common and under which conditions. Further study is needed to assess in situ reproduction of Cibotium in
Hawai'i, and a comparison of island versus continental populations within the family of
Cibotiaceae would be useful to determine if vegetative reproduction may be a unique' evolutionary adaptation to certain insular environmental conditions or if this reproductive strategy is common across the genera of Cibotium.
Conservation Implications
Can and should Cibotium be used in restoration efforts to facilitate natural regeneration of other native species and suppression of invasive vegetation? Correlational results indicate that mature individuals are related to understory vegetation dynamics, and may possibly help to maintain a more native forest. Conversely, mature C. chamissoi may also be restricted to these native areas, and recruitment has been shown to be limited by feral pigs in unprotected areas. Principal components analysis results, though interpreted with caution, nevertheless point to variation in annual rainfall, steeper slopes, and a more dense and native overstory and understory as important variables to C. chamissoi population structure in need of further study. Forest management to encourage a healthy population structure of C. chamissoi appears to require fencing where C. chamissoi is more vulnerable, such as areas of gradual slopes and areas of lower rainfall. In areas of high rainfall, where population structure and abundance appear healthy but invasive vegetation is becoming established, weed control may help to maintain this species. The results obtained in this study cannot answer the question of whether the lack of tree ferns
97 in an area truly indicates declining forest health on O'ahu due to the nature of selective sampling locations to measure population structure. However, as C. chamissoi is vulnerable as a slow-growing, late to mature, and long-lived species, cautionary measures should be taken to protect this species and the forests where it naturally occurs.
98 APPENDIX I. FIELD DATA SHEETS
Date: Site Name: Observers: P\.l.
Site Directions:
GPS point name:
Map of Plot wI Fern # and Canopy Trees
99 Date: Site Name: Elevation: Observers: Pg.
r~- r;--;r.; ' ... '·r·' ""'Sow' ;. • • ",.J ~Mcilsture Slope: Aspect: Drainacie: :'t;poQr~Phv: ! :Cla1.s'i: Substrate: well crest Dry Dry- I Not I' 'It" rl Nalivel Alien Est. Oversto!:X Hei9ht: moderate upper slope Mesic I Fenced . Fenced. circle O. Veg . V~ I poor mid slo e Mesic Wet- hydric lower slope Mesic Degree of Pig Disturbance: high med low none I gulch bottom Wet plateau-flat
Canopy No. of No. of Age Vigor: Trunk Hofiz. Vertical Height Frond Healthy Dying Class: Healthy Basal Top Trunk Trunk from Length: No. of Fronds Fronds No. of Mature No. of Moderate Diam. Diam" Length Length ground longest Fertile (>50% «50% Unfurling Immature No. of Trunk Epiphytes Poor Fern No. Sp. (em) (em) (em) (em) (m) frond (m Fronds green) green) Croziers Sporeling Trunks Buds YIN Dead
"-- _.. -
100 ...... Oat, Site...... , ...... N'."""<;;...... "'<;;, v<;;'''''. p,r [.'i~I~CJ';.... Allen Fenced Not Fenced i~2ne ; .. NativeVeg Veg Overstory Species Understory Species
Estimate Species 0/0 Cover Sp. Sp. Sp. Sp. Pig Activity
Rank piQ activity 0 to 5, 5 beina the most No. of Obs. Aoe of Slon Rank Track: Di in : Browse: Scat:
Estimate total no. of pigs recently in area: Notes:
Understory Plots
f'Plot.1: .. , .... ,,- ~ , Piot2:?P}c',,_ ts1¥1 ·A!'14P!'\2k.., Plot'3!,,:;:f! ",;Wt-..l¥";-,t·,,.~a4' .!f'./$ Plot 4folt~"',ft?\-S,,1~~~'\":-';'l'~~~,:.i<;~ PlotS:,-", 'Io-~ , ",",.1)o";'~~"'i Species 0/0 Cover Species % Cover Species % Cover S ecies D/O Cover S ecies % Covel
101 Date: Site Name: Fenced Not Fenced I~~~I Native Veg I Alien Veg-,
Overstory Density Species North East South West Additional Notes: 1 Average: Trans. # Overstory Distance Density: I(x 1.04-100) Total: Total: Total: Total:
2 Average: Trans. # Overstory Distance Density: I(x 1.04-100) Total: Total: Total: Total:
3 Average: Trans. # Overstory Distance Density: I (x 1.04-100) Total: Total: Total: Total:
4 Average: Trans. # , Overstory Distance Density: (x 1.04-100) Total: Total: Total: Total:
5 Average: Trans. # Overstory Distance Density: (x 1.04-100) Total: Total: Total: Total:
102 ApPENDIX II. OUTPLANTING TRIAL: INDIVIDUAL C. CHAM/SSG/DATA Hapu'u Outplahting
Date: 2/22/06 Site Name: Kahanahaiki Anti Barbara's Observers: _---'-N"A"--______
- Canopy Frond No.-of No. of Vigor: -- - Vertical Height Length: Healthy Dying Spores or Healthy Basal Trunk from longest Fronds Fronds No. of· Pot Size buttons Moderate Overstory Understory Kay's Cire. Length ground frond (>50% « 50% Unfurling at Date 1st Collected Poor Density Density Fern No. No. (em) (em) (em) (em) green) green) Croziers Qutplant Potted From Dead (%) ('Yo) Kapakahi 386 N-15 7 2 18 28 8 0 2 6" 5114/2000 Gulch H 71.4 41
387 N-12 9.5 3.5 37 51 4 0 2 2 gal 11712000 Ohikilolo H 69.06 56
Pisonia 388 N-1 3 2 20 26 8 0 1 6" 11/00/2002 Ohikilolo H 58.66 85 v"~an,,a ~ Area 389 N-7 13 5 68 85 3 0 2 2 aal 3/29/2003 Ridge H 53.98 80 button- 390 N-5 10 5 34 52 5 0 2 1 cal 4/9/2003 Kahanahaiki H 66.2 35
391 N-2 10.5 4 34 58 7 0 2 1 gal 11/30/2002 Ohikilolo H 56.25 105 button- 392 N-11 17 3.5 33 73 4 0 2 2 gal ? Kahanahaikl H 49.82 25 ...... ¥l' ."",<" ;~ •... -!i #.- ~"-- ~ .~:;. -v.:~r; t fi';w" :" ' ."C . .. l ... : J . 1•. """" . I'~J' 'lir')'~,,~ ~.--- - r-'J [";" f .' 1"'-' "-',.:~~~~: ,':if . "".t,..,. ~ ., ,+,. rY .•...r4< ··· ) ·tt' :~, t~~~ _.0 -~. ,;,,y ,.- If.t~4 . - _.,. I~J L" .. ,' r""-''! ~,,-'--., ~i
393 N-13 7 3.5 ' 16 37 7 0 2 2 gal 11712000 Ohikilolo H 49.04 15 button- 394 N-9 14 7.5 54 64 3 1 2 2 gal 4/29/2005 Kahanahaiki H 53.72 1
395 N-14 8 1.5 17 31 7 0 1 6" 117/2000 Ohikilolo H 71.14 65 Kapakahi Dead 396 N-16 1 9 21 6 1 2 6" 5/14/2000 Gulch H 68.02 27 Kapakahi Guava 397 N-17 7 3 17 38 8 1 2 1 aal 5/14/2000 Gulch H 63.6 20
Area 398 N-3 10 3.5 27 56 7 0 1 1 gal 11/30/2002 Ohikilolo H 49.3 50
399 N-6 9 4 38 42 7 0 2 1 aa' 1/7/2000 Ohikilolo H 38.9 2 WaahiJa 400 N-8 13 6 33 69 3 0 1 2 gal 3/2912003 Ridge H 51.12 20
103 ApPENDIX III. RESULTS OF SOIL ANALYSES: PAIRED PLOTS
700 ii2'__ 600 § 500 [ ';; 400 ::1300 ~ I'i l. 200 " -r-> --Ii" = 100 h;::r-j)" fti---' -l l;! h 0 ~ 0 : 1:"1 ~1:"1 ~ I~~:J In 1"1 : 1°':1 Ohikilolo 3 Points Kabanahaik' Pahole 1 Fahole 2,
Plot PI"
3500 3' 3000 e 2500 .:, 2000 1500 ~ u " 1000 ~ 500 RJ=t~"-~ - -" ~ o. , ,'" TI ilif ~: 1'1 ~ In IOllt I, 1o,: II' 10" lin Io,~~ 10 I0" Ohlk11olo 3 POlfltS [Kahanahai Pahok 1 Pahole 2 PI", PIo,
be 1400 e 1200 1000 " " • 800 600 400 ~ [:;::;:::f; 200 f _,' ;n~T '"~ ::m=lj: '" ~KH'Hffi 00 0 To lOut 10 10"1 :hl~:: 1;1':'::' IK:"~~,~:J Pahole 1 Pahole 2 PIo, PI", e; 60 50 l , 40 ! " 30 20 1 10 ] 0 I, 10"1 In 10,,11, 10"1 In 1°"11, 1°"1 Ohikilolo 3 Puint' lK:r.hanahalki Pahale 1 Pahole 2 PIO[
104 Soil Phosphorous (P) Soli Nitrogen (N) Soil OrglUlk Carbon (0 Soil pH 0000 __ ..... N W .,. _..... """" 0"""'0."" 000 C C ON:;..O'OO ..... N~ 0'-"0'-"0'-" ------n!i.ilCilruo --ohiki!olo '00 ',,--~- Weed' ". Weed ~«dy_ ~y--- Ohikilolo Obikilolo ~ Ohikilolo " "01,+' Ohikilolo Ri ana aJ 1 Karu.nahalkl Ka .. OJ aJ KahaoahaJki , '7; Stee ~ Steeo ---.it~__ € s,~ ~ ··c ~ ;;~ Klthaoahaiki Klthanahaiki ;>-'-" ~- Kahanahaiki Kahal1.lh.iki ~- 1- , ~ 3 Points ~ 31'oiots ~ 3 Points I¥ ~ 3 Poim, ~ ~ ~ ~ l'ahule 2 ~ Pahole 2 ~ Pahole 2 Pahole 2 .. [ ..;: '"C ~ ;- Pahole 1 Pahole I l'ahule 1 Pahole 1 t; - en
---Ly61l-- Lyoll ~~l'! Lyon ~ Arboutum Arbotetum Arboretum Arboretum (J) 0 " "'- " Aie
OhikiJolo ~ ------ufiitiIOlo , '++++ -----U1iil Lyun .YO" --L-YO;;-- '" Atboretum Arboretum 6: Arbor<:tum Aiea Ridge Aiea Ridge: ~ Aiea Ridge o~ ~oa ~ ----...;;;;;;- ~~11S"i'ia ~ ~ Valley Valle V~lle ~ ~ Kahuku ~. K~huku " K3huku LITERATURE CITED Army Natural Resource Center. 2004. U.S. Army Garrison Hawai'i O'ahu Training Areas Natural Resource Management Final Report. Pacific Cooperative Studies Unit. Alyokhin, A. V., P. Yang and R. H. Messing. 2001. Distribution and parasitism of Sophonia rufofascia (Homoptera: Cicadellidae) eggs in Hawai'i. Annals of the Entomological Society ofAmerica 94: 664-669. Alyokhin, A. V., P. Yang and R. H. Messing. 2004. Oviposition of the invasive two spotted leafhopper on an endemic tree: effects of an alien weed, foliar pubescence, and habitat humidity. Journal ofInsect Science 4: 13. Ash, J. 1987. Demography ofCyathea hornei (Cyatheaceae), a tropical tree-fern in Fiji. Australian Journal ofBotany 35: 331-342. Asner, G. P. and P. M. Vitousek. 2005. Remote analysis of biological invasion and biochemical change. Proceedings of the National Academy ofSciences 102: 4383-4386. Bartlett, M. S. 1954. A note on the multiplying factors for various chi square approximations. Journal of the Royal Statistical Society. Series B (Methodological) 16: 296-298. 106 Becker, R. E. 1976. The phytosociological position of tree ferns (Cibotium Spp.) in the montane rain forests on the Island ofHawai'i. Dissertation for the degree of Doctor of Philosophy. Botanical Sciences, University ofHawai'i at Manoa. Bernabe, N., G. Williams-Linera and M. Palacios-Rios. 1999. Tree ferns in the interior and at the edge of a Mexican cloud forest remnant: spore germination and sporophyte survival and establishment. Biotropica 31: 83-88. Bittner, J. and S. W. Breckle. 1995. The growth rate and age of tree fern trunks in relation to habitats. American Fern Journal 85: 37-42. Brown, J. H. and M. V. Lomolino. 1998. Biogeography. 2nd ed., Sinauer Associates Inc., Sunderland, MA. Buck, M. G. 1982. Hawaiian treefern harvesting affects forest regeneration and plant succession. Pacific Southwest Forest and Range Experiment Station, United States Department of Agriculture Forest Service: research note PWS-355, Berkeley, CA. Cabin, R. J., S. G. Weller, D. H. Lorence, T. W. Flynn, A. K. Sakai, D. Sandquist and 1. J. Hadway. 2000. Effects of long-term ungulate exclusion and recent alien species control on the preservation and restoration of a Hawaiian dry tropical forest. Conservation Biology 14: 439-453. Callaway, R. M. and J. 1. Maron. 2006. What have exotic plant invasions taught us over the past 20 years? Trends in Ecology & Evolution 21: 369-374. 107 Carl quist, S. 1974. Island Biology. Columbia University Press, New York and London. Catell, R. B. 1966. The scree test for the number offactors. Multivariate Behavioral Research 1: 245-276. Christopherson, R. W. 2004. Elemental Geosystems. 4th ed., Pearson Education, Inc., New Jersey. Conant, D. S. 1978. A radioisotope technique to measure spore dispersal of the tree fern Cyathea arborea Sm. Pollen & Spores 20: 583-593. Conant, D. S., D. S. Stein, A. C. Valinski, P. Sudarsanam and M. E. Ahearn. 1994. Phylogenetic implications of chloroplast DNA variation in the Cyatheaceae. 1. Systematic Botany 19: 60--:72. Cooray, R. G. 1974. Stand structure ofa montane rain forest on Mauna Loa, Hawai'i. Department of Botany, University ofHawai'i: Technical Report No. 44: U.S. International Biological Program, Island Ecosystems IRP, Honolulu. Cuddihy, L. W. and C. P. Stone. 1990. Alteration ofNative Hawaiian Vegetation: Effects ofHumans, Their Activities and Introductions. University of Hawai'i, Cooperative National Park Resources Studies Unit, Honolulu, Hawai'i. Culliney, T. W. 1998. Site of oviposition and description of eggs of Sophonia rufofascia (Homoptera: Cicadellidae: Nirvaninae), a polyphagous pest in Hawai'i. Proceedings-Hawaiian Entomological Society 33: 67-73. 108 D'Antonio, C. M. and T. L. Dudley. 1995. Biological invasions as agents of change on islands versus mainlands. In P. Vitousek, L. Loope and H. Andsersen (Eds.), Biological Diversity and Ecosystem Function on Islands. Springer-Verlag, New York: pp. 123-134. Daehler, C. 2005. Upper-montane plant invasions in the Hawaiian Islands: patterns and opportunities. Perspectives in Plant Ecology, Evolution and Systematics 7: 203- 216. Diong, C. H. 1982. Population biology and management of the feral pig (Sus scrofa L.) in Kipahulu Valley, Maui. Dissertation for the degree of Doctor of Philosophy. Department of Zoology, University ofHawai'i at Manoa. Drake, D. R. and D. Mueller-Dombois. 1993. Population development of rain forest trees on a chronosequence of Hawaiian lava flows. Ecology 74: 1012-1019. Durand, L. Z. and G. Goldstein. 200 I. Growth, leaf characteristics, and spore production in native and invasive tree ferns in Hawai'i. American Fern 10urna191: 25-35. Dyer, A. F. 1979. The culture offern gametophytes for experimental investigation. In A. F. Dyer (Ed.) The Experimental Biology ofFerns. Academic Press, New York: pp.253-305. Fosberg, F. R. 1942. Uses of Hawaiian Ferns. American Fern 10urna132: 15-23. 109 Friend, D. 1. C. 1974. Shade adaptation of the Hawaiian tree-fern (Cibotium glaucum (Sm.) H. & A.). Department of Botany, University ofHawai'i: Technical Report No. 41: U.S. International Biological Program, Island Ecosystems IRP, Honolulu. Fukada, M. T. 1996. Distribution, host range, and seasonal abundance ofthe twospotted leafhopper, Sophonia rufofasica (Kuoh and Kuoh) in Hawai'i. M.S. Thesis. University of Hawai'i at Manoa. Honolulu. Funk, V. A. and W. L. Wagner. 1995. Biogeographic patterns in the Hawaiian Islands. In W. L. Wagner and V. A. Funk (Eds.), Hawaiian Biogeography: Evolution on a Hot Spot Archipelago. Smithsonian Institution Press, Washington, D.C.: pp. 379- 419. Gastony, G. J. 1981. Spore morphology in the Dicksoniaceae. 1. The genera Cystodium, Thyrsopteris, and Culcita. American Journal ofBotany 68: 808-819. Gastony, G. J. 1982. Spore morphology in the Dicksoniaceae. II. The genus Cibotium. Canadian Journal ofBotany 60: 955-972. Giambelluca, T. W. 1983. Water balance of the Pearl Harbor-Honolulu basin, Hawai'i, 1946-1975. Water Resources Research Center, University ofHawai'i: Technical Report 151, Honolulu. Giambelluca, T. W., M. A. Nullet and T. A. Schroeder. 1986. Hawai 'i Rainfall Atlas. Hawai'i Division of Water and Land Development, Department of Land and Natural Resources: Report R76, Honolulu. 110 Heu, R. and B. Kumashiro. 1989. Notes & exhibitions: Sophonia rufofasica. Proceedings-Hawaiian Entomological Society 29: 16-17. Johnson, M. T., P. Yang, J. T. Huber and V. P. Jones. 2001. Egg parasitoids of Sophonia rufofascia (Homoptera: Cicadellidae) in Hawai'i Volcanoes National Park. Biological Control 22: 9-15. Jones, V. P., P. Anderson-Wong, P. A. Follett, P. Yang, D. M. Westcot, J. S. Hu and D. E. Ullman. 2000. Feeding damage of the introduced leafhopper Sophonia rufofascia (Homoptera: Cicadellidae) to plants in forests and watersheds of the Hawaiian Islands. Environmental Entomology 29: 171-180. Kaiser, H. F. 1974. An index of factorial simplicity. Psychometrika 39: 31-35. Kaneshiro, K. Y. 1989. Uniqueness ofHawai'i's biota. In C. P. Stone and D. Stone (Eds.), Conservation biology in Hawai'i. Cooperative National Park Resources Studies Unit, University of Hawai'i at Manoa, Honolulu: pp. 7-10. Korall, P., K. M. Pryer, J. S. Metzgar, H. Schneider and D. S. Conant. 2006. Tree ferns: monophyletic groups and their relationships as revealed by four protein-coding plastid loci. Molecular Phylogenetics and Evolution 39: 830-845. Lindenmayer, D. B., R. B. Cunningham, C. F. Donnelly and J. F. Franklin. 2000. Structural features of old-growth Australian montane ash forests. Forest Ecology and Management 134: 189-204. III Livingston, D. N. 1992. The Geographical Tradition. Blackwell, Oxford. Loh, R. K. H. 2004. Early plant establishmentfollowing experimental removal of invasive Morellafaya (Ait.) Wilbur stands in a Hawaiian forest. Dissertation for the degree of Doctor of Philosophy. Botanical Sciences, University ofHawai'i at Manoa. Loope, L. L. and P. G. Scowcroft. 1985. Vegetation response within exclosures in Hawai'i: a review. In C. P. Stone and 1. M. Scott (Eds.), Hawai'i's Terrestrial Ecosystems: Preservation and Management. University ofHawai'i Press, Honolulu: pp. 377-402. Loope, L. L. and D. Mueller-Dombois. 1989. Characteristics of invaded islands, with special reference to Hawai'i. In J. A. Drake, H. A. Mooney, F. d. Castri, R. H. Groves, F. H. Kruger, M. Rejmanek and M. Williamson (Eds.), Biological Invasions: A Global Perspective. John Wiley & Sons, New York: pp. 257-280. MacCaughey, V. 1916. The tree ferns of Hawai'i. The American Botanist 1: 1-9. MacCaughey, V. 1918. An ecological survey of the Hawaiian pteridophytes. The JOurnal ofEcology 6: 199-219. Medeiros, A. C., L. L. Loope, T. Flynn, S. J. Anderson, L. W. Cuddihy and K. A. Wilson. 1992. Notes on the status of an invasive Australian tree fern (Cyathea cooper!) in Hawaiian rain forests. American Fern Journal 82: 27-33. 112 Medeiros, A. C., 1. 1. Loope and S. 1. Anderson. 1993. Differential colonization by epiphytes on native (Cibotium spp.) and alien (Cyathea cooperi) tree ferns in a Hawaiian rain forest. Selbyana 14: 71-74. Mueller-Dombois, D. 1973. A non-adapted vegetation interferes with water removal in a tropical rain forest area in Hawai'i. Tropical Ecology 14: 1-18. Mueller-Dombois, D. 1981. Island ecosystems: what is unique about their ecology? In D. Mueller-Dombois, K. W. Bridges and H. 1. Carson (Eds.), Island Ecosystems: Biological Organization in Selected Hawaiian Communities. US/lBP Synthesis Series 15. Hutchinson Ross Publishing Co., Stroudsburg and Woods Hole: pp. 485-50l. Mueller-Dombois, D., R. G. Cooray, J. E. Maka, G. Spatz, W. C. Gagne, F. G. Howarth, 1.1. Gressitt, G. A. Samuelson, S. Conant and P. Q. Tomich. 1981. Structural variation of organism groups studied in the Kilauea Forest. In D. Mueller Dombois, K. W. Bridges and H. 1. Carson (Eds.), Island Ecosystems: Biological Organization in Selected Hawaiian Communities. US/IBP Synthesis Series 15. Hutchinson Ross Publishing Co., Stroudsburg and Woods Hole: pp. 231-317. Mueller-Dombois, D. 2005. A silvicultural approach to restoration of native Hawaiian rainforests. Lyonia 8: 59-63. Mueller-Dombois, D. In press. The Hawaiian ahupua'a land use system: its biological resource zones and the challenge for silvicultural restoration. In N. 1. Evenhuis 113 and 1. M. Fitzsimmons (Eds.), Biology ofHawaiian Streams and Estuaries. Bishop Museum Bulletin in Cultural and Environmental Studies, vol. 3: pp. 23- 33. Nelson, R. E. and E. M. Homibrook. 1962. Commercial uses and volume of Hawaiian tree fern. U.S. Forest Service. Pacific Southwest Forest and Range Experiment Station: Technical Paper 73, Berkeley, CA. Ostertag, R. and 1. H. Verville. 2002. Fertilization with nitrogen and phosphorus increases abundance of non-native species in Hawaiian montane forests. Plant Ecology 162: 77-90. Ough,K. 2001. Regeneration of wet forest flora a decade after clear-felling or wildfire is there a difference? Australian Journal ofBotany 49: 645-664. Ough, K. and A. Murphy. 2004. Decline in tree-fern abundance after c1earfell harvesting. Forest Ecology and Management 199: 153-163. Page, C. N. and P. 1. Brownsey. 1986. Tree-fern skirts: a defense against climbers and large epiphytes. Journal ofEcology 74: 787-796. Palmer, D. 1994. The Hawaiian Species of Cibotium. American Fern Journal 84: 73-85. Palmer, D. 2003. Hawai 'j's Ferns and Fern Allies. University ofHawai'i Press, Honolulu. 114 Ripperton,1. C. 1924. The Hawaiian tree fern as a commercial source of starch. Hawai'i Agricultural Experiment Station Bulletin: 53. U.S. Govt. Print Office, Washington. Sakai, A. K., W. L. Wagner and L. A. Mehrhoff. 2002. Patterns of endangennent in the Hawaiian flora. Systematic Biology 51: 276-302. Santiago, L. S. 2000. Use of coarse woody debris by the plant community ofa Hawaiian montane cloud forest. Biotropica 32: 633--641. Schmitt, J. L. and P. G. Windisch. 2006. Phenological aspects of frond production in Alsophila setosa (Cyatheaceae: Pteridophyta) in southern Brazil. The Fern Gazette 17: 263-270. Schofield, E. K. 1989. Effects of introduced plants and animals on island vegetation: examples from the Galapagos Archipelago. Conservation Biology 3: 227-238. Scowcroft, P. G. 1997. Mass nutrient dynamics of decaying litter from Passiflora mollissima and selected native species in a Hawaiian montane rain forest. Journal o/Tropical Ecology 13: 407-426. Sheffield, E. 1996. From pteridophyte spore to sporophyte in the natural environment. In J. M. Camus, M. Gibby and R. J. Johns (Eds.), Pteridology in Perspective. Kew, Royal Botanic Gardens, pp. 541-549. 115 Smith, A. R., K. M. Pryer, E. Schuettpelz, P. Koral!, H. Schneider and P. G. Wolf. 2006. A classification for extant ferns. Taxon 55: 705-731. Smith, C. W. 1985. Impact of alien plants on Hawai'i's native biota. In C. P. Stone and 1. M. Scott (Eds.), Hawai 'i's Terrestrial Ecosystems: Preservation and Management. University ofHawai'i Press, Honolulu: pp. 180-250. Spatz, G. and D. Mueller-Dombois. 1973. The influence offeral goats on koa tree reproduction in Hawai'i Volcanoes National Park. Ecology 54: 870-876. Stone, C. P., L. W. Cuddihy and 1. T. Tunison. 1992. Responses of Hawaiian ecosystems to removal of feral pigs and goats. In C. P. Stone, C. W. Smith and 1. T. Tunison (Eds.), Alien Plant Invasions in Native Ecosystems ofHawai 'i: Management and Research. University of Hawai'i Press, Honolulu: pp. 666-704. Vitousek, P. M. 1988. Diversity and biological invasions of oceanic islands. In E. O. Wilson (Ed.) Biodiversity. National Academy Press, Washington, D.C.: pp. 181- 189. Vitousek, P. M. and L. R. Walker. 1989. Biological invasion by Myricafaya in Hawai'i: plant demography, nitrogen fixation, and ecosystem effects. Ecological Monographs 59: 247-265. Vitousek, P. M. 1992. Effects of alien plants on native ecosystems. In C. P. Stone, C. W. Smith and 1. T. Tunison (Eds.), Alien Plant Invasions in Native Ecosystems of 116 Hawai'i: Management and Research. University ofHawai'i Press, Honolulu: pp. 29-41. Vitousek, P. M., G. Gerrish, D. R. Turner, L. R. Walker and D. Mueller-Dombois. 1995. Litterfall and nutrient cycling in four Hawaiian montane rainforests. Journal 0/ Tropical Ecology 11: 189-203. Vtorov,1. P. 1993. Feral pig removal: effects on soil microarthropods in a Hawaiian rain forest. Journal o/Wildlife Management 57: 875-880. VuJcz, R., L. VuJcz, L. D. Greer and G. Lawrence. 2002. Sustainable production of tree ferns in South-East Australia. The Fern Gazette 16388-392. Wagner, W. L., D. R. Herbst and R. S. N. Vee. 1985. Status of the native flowering plants of the Hawaiian Islands. In C. P. Stone and 1. M. Scott (Eds.), HawaiTs Terrestrial Ecosystems: Preservation and Management. University ofHawai'i Press, Honolulu: pp. 23-74. Wagner, W. L., D. R. Herbst and S. H. Sohmer. 1990. Manual o/the Flowering Plants 0/ Hawai'i. Revised Edition, University of Hawai 'i Press and Bishop Museum Press, Honolulu. Walker, L. R. and G. H. Aplet. 1994. Growth and fertilization responses of Hawaiian tree ferns. Biotropica 26: 378-383. Wardlaw, A. C. 2002. Conservation of tree ferns ex situ. The Fern Gazette 16: 393-397. 117 Wick, H. L. and G. T. Hashimoto. 1971. Frond development and stem growth of tree-fern in Hawai'i. Pacific Southwest Forest and Range Experiment Station, U.S. Forest Service: Research note PSW-237, Berkeley, CA. Zhang, x.-C., J.-S. Jia and G. M. Zhang. 2002. Survey and evaluation of the natural resources of Cibotium barometz (L.) J.Smith in China, with reference to the implementation ofthe CITES convention. The Fern Gazette 16: 383-392. 118