POLLINATION ECOLOGY OF HAWAIIAN COASTAL PLANTS

A THESIS SUBMITTED TO THE GRADUATE DIVISION OF THE UNIVERSITY OF HAWAI‘I AT MĀNOA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF

MASTER OF SCIENCE IN BOTANY (ECOLOGY, EVOLUTION AND CONSERVATION BIOLOGY) AUGUST 2014

By Kimberly R. Shay

Thesis Committee: Donald R. Drake, Chairperson Kasey E. Barton Andrew D. Taylor

Keywords: pollination network, coastal pollinators, coastal plants, Jacquemontia sandwicensis, Hylaeus

ACKNOWLEDGEMENTS

I would like to thank and acknowledge the following people for all of their help and support throughout this research and my thesis. In particular, I would like to thank my advisor Don Drake for truly being a great mentor. I am tremendously grateful to my committee members, Kasey Barton and Andy Taylor, Heather Sahli for sharing this research with me, the Botany Department faculty and staff, my family and friends for their continued support, and especially my husband, Chris.

Chapter 1: I would like to thank Patrick Aldrich, Melody Euaparadorn, Michelle Elmore, Doug Powless, Jennifer Imamura, Paul Krushelnycky, Stephanie Joe, Will Haines, Jennifer Bufford, Jesse Eiben, Tad Fukami, and Christian Torres-Santana for collecting the data, Luc Leblanc for identifying and managing all of the collected insects, and the State of Hawaiʽi Department of Land and Natural Resources Division of Forestry and Wildlife and Charmian Dang for overseeing the permits to complete this work.

Chapter 2: I would like to thank Kobey Togikawa for all of his hard work, Amelia Harvey for assisting with observations, Christopher Shay, Joshua Boothby, and Hiroshi Park for collecting cuttings, Sheldon Plentovich, Paul Krushelnycky, and Karl Magnacca for identifying insects, and State of Hawaiʽi Department of Land and Natural Resources Division of Forestry and Wildlife and Cynthia King for overseeing the permits to collect the insects. Funding was generously provided by the Hawai‘i Audubon Society, the Charles H. Lamoureux Fellowship from the Botany Department, and the Torrey-Degener-Rock Scholarship from the Botany Department.

ii

TABLE OF CONTENTS

ACKNOWLEDGEMENTS ...... ii

LIST OF TABLES ...... v

LIST OF FIGURES ...... vi

CHAPTER 1. ALIEN INSECTS DOMINATE THE PLANT-POLLINATOR NETWORK OF A HAWAIIAN COASTAL ECOSYSTEM ...... 1

Abstract ...... 1

Introduction ...... 1

Methods...... 4

Results ...... 11

Discussion ...... 22

Conclusion ...... 29

CHAPTER 2. CHAPTER II. THE POLLINATION BIOLOGY OF Jacquemontia sandwicensis (CONVOLVULACEAE) ...... 31

Abstract ...... 31

Introduction ...... 31

Methods...... 34

Results ...... 41

Discussion ...... 47

Conclusion ...... 53

iii

APPENDIX A. NETWORK AND GROUP INDICES OF KAʽENA POINT ...... 55

APPENDIX B. 2009 QUANTITATIVE MATRIX OF KAʽENA POINT ...... 59

LITERATURE CITED ...... 60

iv

LIST OF TABLES

Table 1.1. Plant species in the networks at Ka‘ena Point, O‘ahu, Hawai‘i ...... 7

Table 1.2. Pollinator taxa in the networks at Ka‘ena Point, O‘ahu, Hawai‘i ...... 9

Table 1.3. Network Metrics of the binary and quantitative networks of Ka‘ena Point, O‘ahu, Hawai‘i ...... 14 Table 1.4. Network Indices for the binary and quantitative datasets of Ka‘ena Point, O‘ahu, Hawai‘i ...... 15 Table 1.5. Insect orders represented in networks of Ka‘ena Point, O‘ahu, Hawai‘i ...... 17

Table 1.6. Plant species’ proportions of visits from Hylaeus spp. from the quantitative network of Ka‘ena Point, O‘ahu, Hawai‘i ...... 19 Table 2.1. Floral morphology of Jacquemontia sandwicensis ...... 36

Table 2.2. Treatments in the breeding experiment of Jacquemontia sandwicensis ...... 39

Table 2.3. Pollinators at field sites and their proportions of visits to Jacquemontia sandwicensis ...... 46

Table 2.4. Summary of flowering species at sites ...... 48

v

LIST OF FIGURES

Figure 1.1. Ka‘ena Point, O‘ahu, Hawai‘i ...... 5

Figure 1.2. 2008-2009 Binary plant-pollinator network of Ka‘ena Point, O‘ahu, Hawai‘i ...... 12

Figure 1.3. 2009 Quantitative plant-pollinator network of Ka‘ena Point, O‘ahu, Hawai‘i ...... 13

Figure 1.4. The distribution of visits of generalized pollinators to plant species from the quantitative network of Ka‘ena Point, O‘ahu, Hawai‘i ...... 20 Figure 1.5. The proportion of visits by pollinator taxa out of the total visits in the quantitative network of Ka‘ena Point, O‘ahu, Hawai‘i ...... 21 Figure 1.6. The proportion of visits by plant species out of the total visits of the quantitative network of Ka‘ena Point, O‘ahu, Hawai‘i ...... 23 Figure 1.7. Categories of plant species or pollinator taxa d’ values from the quantitative network of Ka‘ena Point, O‘ahu, Hawai‘i ...... 24 Figure 2.1. Map of Study sites ...... 35

Figure 2.2. Flower of Jacquemontia sandwicensis ...... 37

Figure 2.3. Lab breeding experiment ...... 43

Figure 2.4. Field breeding experiment ...... 44

Figure 2.5. Control treatment (no manipulation, open to natural pollinators) at 3 sites and results ...... 45

vi

CHAPTER 1. ALIEN INSECTS DOMINATE THE PLANT-POLLINATOR NETWORK OF A HAWAIIAN COASTAL ECOSYSTEM

ABSTRACT

Little is known regarding pollination webs involving island coastal plants and pollinators, and the roles that non-native flower visitors may play in these interaction networks. Plant- pollinator observations collected in March 2008 and 2009 were used to describe the pollination network for Kaʽena Point, one of Hawaii’s best-conserved coastal communities. The network includes 15 native plant species, 2 native insect species, and 27 non-native insect taxa, forming 119 interactions. Network connectance is 29.4% and weighted nestedness is 18.3, which is similar to other dry habitat, island networks. The network’s structure has a compartment of generalized pollinators followed by more specialized pollinators. Nearly all plant species interact with two or more generalist pollinators and a variable number of specialists. Small, non-native bees (Lassioglossum, Ceratina), wasps (Proconura), and flies (Tachinidae) were responsible for 71.6% of the flower visits and visit five plant species not visited by native bees. The two native visitors, Hylaeus anthracinus and H. longiceps, (both proposed as threatened/endangered) belong to a radiation of > 60 endemic species in Hawaii’s only genus of native bees. Hylaeus spp. (especially females) provided 20.6% of the flower visits, foraging on many species and at high frequencies, including the endangered Scaevola coriacea and Sesbania tomentosa. In Hawaii’s coastal habitat, non-native insects form novel interactions with native species, and can maintain an ecosystem’s function after losing most of the original native species. However, the two remaining native Hylaeus species are still important pollinators to many native plants on which they rely for nectar and pollen resources.

INTRODUCTION

Species interaction networks have been referred to as “the architecture of biodiversity” because of their importance for community composition and dynamics (Jordano et al 2006, Bascompte and Jordano 2007). Network analyses are increasingly being used to understand how communities are changing as a result of species invasions or extinctions (Memmott et al 2004, 1

Bjerknes et al 2007, Kaiser-Bunbury et al 2010a). Plant-pollinator networks are especially important, because 87.5% of the world’s flowering plant species are pollinated by animals, particularly insects (Ollerton 2011), and networks are a way to assess a community’s pollination activity, importance for community composition, and dynamics. A pollination network can be represented either as a matrix or as a two-mode data set with species as nodes and the interactions between species as links (Jordano 1987, Jordano et al 2006). Networks may be described qualitatively, depicting simply the presence or absence of interactions, or quantitatively, based on the frequency of visits within an interaction. The interactions and visits can be used to calculate network indices that are then used to interpret patterns and processes in the community (Dormann et al 2009). For example, networks can be analyzed to identify keystone species (Christianou and Ebenman 2005), monitor biodiversity (Forup et al 2008), and determine the functional roles of non-native species (Kaiser-Bunbury et al 2010b, Tylianakis et al 2010).

Plant-pollinator networks on islands are of interest for a variety of reasons. Owing to their isolation, island biotas are often naturally disharmonic and relatively species-poor (MacArthur and Wilson 1967, Whittaker and Fernandez-Palacios 2007). As a result, their plant- pollinator networks commonly have generalized mutualisms (most species interact with many mutualists; Olesen 2002, Dupont 2003) and taxonomic disharmony in the pollinators (Kaiser- Bunbury et al 2010b). This often translates into small network sizes, high connectance (= a high proportion of potential plant-pollinator interactions being realized), and a super-generalist species which dominates interactions (Olesen 2002, Dupont et al 2003, Trojelsgaard and Olesen 2012). Additionally, many island systems involve native species that have become rare (or gone extinct), and one or more non-native species dominating interactions or visits (Dupont et al 2003, Kaiser-Bunbury et al 2010b, Traveset et al 2013). Non-native species typically enter a network by interacting with generalized native species. This has been documented with non-native plant invasions (Bjerkens et al 2007, Bartomeus et al 2008, Traveset and Richardson 2006) and the honey bee, Apis mellifera (Dupont et al 2003, Kaiser-Bunbury 2011). Once in the network, non- native species have a large impact by interacting with many partners (Bartomeus et al 2008,

2

Padron et al 2009) and affecting other species’ interactions (Aizen et al 2008, Kaiser-Bunbury et al 2011, Traveset et al 2013).

In Hawai‘i, few studies have assessed relationships between plants and their insect pollinators (Howarth 1985, Medeiros et al 2013). Hawai‘i’s native insect fauna is disharmonic because many of the taxa of pollinators common in other ecosystems are underrepresented or absent here. For example, social insects (Wilson 1990) and butterflies (Hawai‘i has only 2 endemic butterflies; Zimmerman 1958). Hawai‘i’s only bees are a radiation of >60 species of Hylaeus, (Colletidae, yellow-faced bees) (Magnacca 2007). They have been found visiting flowers of native plants throughout Hawai‘i (Magnacca 2007, Wilson et al 2010, Koch and Sahli 2013, Krushelnycky 2014). Unfortunately, lowland native insect communities have been devastated by a variety of factors, including interactions with non-native insects (Howarth 1985, Cole et al 1992), novel predators (i.e. Kraus et al 2012), and habitat loss (Asquith and Messing 1993). As a result of these disruptions, many insect pollinators, including many Hylaeus spp., are rare or extinct (Howarth 1985, Cox 2000, Magnacca and King 2013).

Little is known about the coastal strand ecosystem of Hawaiʽi beyond general descriptions of vegetation and the physical environment (Richmond and Mueller-Dombois 1972, Gagne and Cuddihy 1990, Warshauer et al 2009), and the observation that these ecosystems have been severely impacted by development, recreation, and non-native species (Richmond and Mueller-Dombois 1972, Cuddihy and Stone 1990, Warshauer et al 2009). Three coastal plant species require a pollinator for outcrossing or for transfer of self-pollen to the stigma for seed set (Hopper 2002, Yorkston and Daehler 2006, Shay Chapter 2). Observations of flower-visitors of five other coastal plant species have recorded 2 native and at least 19 non-native insect taxa, mostly bees, ranging from likely pollinators to nectar thieves (Hopper 2002, Elmore 2008, Pleasants and Wendel 2010, Shay Chapter 2). These studies’ authors frequently postulate the importance of native Hylaeus species as likely pollinators and in some cases, dominant visitors.

Threats such as habitat loss, fragmentation, and non-native species disrupt plant- pollinator interactions on islands around the world (Traveset and Richardson 2006, Kiers et al

3

2010, Potts et al 2010). The startling numbers of non-native insects (>2500 species) and plants (>900 species) established in Hawai‘i (Pimental et al 2005) and recent observations of non- native flower visitors on native coastal plants (cited above), suggest that native plant-pollinator interactions there may be affected by non-native species (Traveset and Richardson 2006, Kaiser- Bunbury et al 2010b). The alteration or loss of plant-pollinator interactions may impact the fitness of the corresponding partner (Kiers et al 2010, Potts et al 2010, Colwell 2012). Based on the lack of information on native plant-pollinator interactions, the presence of endangered and rare native species in the coastal community, and the high numbers of invasive species there, coastal pollination webs warrant further investigation.

The objectives of this research are to 1) describe a plant-pollinator network qualitatively and quantitatively, in a native community of coastal plants, and 2) determine the relative diversity and potential importance of native and alien insects in the network. I predict that 1) the network will include both native and non-native pollinators, and 2) non-native pollinators will interact with many partners and contribute a large proportion of visits. I expect our network to have similar parameters and results to other small networks of islands, which typically have fewer species with higher proportion of partners compared to continental networks (Trojelsgaard et al 2013).

METHODS

Study site - The study was carried out in late March of 2008 and 2009 at Ka‘ena Point Natural Area Reserve on the northwest coast of O‘ahu (Figure 1). Ka‘ena Point has a low annual rainfall of approximately 650 mm / year, concentrated mainly in the winter with drought in the summer (Richmond and Mueller-Dombois 1972, Gagne and Cuddihy 1990, Giambelluca et al 2013), and mean annual temperature 22.2°C to 25.5°C (Richmond and Mueller-Dombois 1972, Giambelluca et al 2014). This site area also experiences solar radiation, constant and strong winds, and effects of salt spray (Richmond and Mueller-Dombois, 1972, Gagne and Cuddihy 1990, Giambelluca et al 2014). The study area contains protected native coastal strand and dune vegetation dominated by shrubs and low growing plants.

4

A

B C D

Figure 1. Ka‘ena Point, O‘ahu, Hawai‘i. (A) Approximate transect through coastal habitat at Ka‘ena Point (red dashed line). (B) Example of dune vegetation, including Sesbania tomentosa and Scaevola taccada. (C) Example of coastal strand vegetation. (D) Map of O‘ahu, Hawai‘i with Ka‘ena Point marked in the northwest point of the island (yellow arrow).

5

Observations – Observations were collected previous to this thesis in 2008 and 2009. Ten observation points were placed 20-50 m in from the shoreline, at approximately 100 m intervals from east to west along the north edge of the Natural Area Reserve (Figure 1). Alternate observation points were observed on every other day for 7 days in 2008 and 4 days in 2009.

Fifteen native plant species and no non-native species (other than grasses) flowered regularly during observations (Table 1). At each observation point, a representative of each plant species within 10 m of the observation point was randomly selected and observed for 15 min, if possible. For a plant species not occurring within 10 m of the observation point, the closest flowering individual within 30 m was observed or in some cases at a different observation point to obtain similar amounts of observation times for each species. Observations occurred from 9:30 a.m. to 5:00 p.m., and totaled 72 hr and 30 min in 2008 and 56 hr in 2009. Observers recorded: observation point, plant species, number of flowers observed, flower visitor, number of flowers visited by the visitor, duration of visit, if reproductive structures of the flower were touched by the visitor, and if the visitor foraged for nectar, pollen, or both. The flower visitors were identified in the field when possible; when insects could not be identified in the field they were collected, when possible, for later. Collected insects were deposited, identified and resolved to the lowest possible taxonomic level by Dr. Luc Leblanc of University of Hawai‘i at Mānoa’s Insect Museum.

In 2009, flower abundance (no./m2) was determined in a 20 x 20 m plot centered on each observation point. For species having few or large flowers, all flowers in the plot were counted. For species having small or numerous flowers, flowers were counted in five, parallel 0.5 x 20 m transects spaced 5 m apart. For plant species having small flowers borne in specialized inflorescences (Boerhavia, Melanthera), each inflorescence was counted as a single flower.

Network analysis – As part of my master’s thesis, I am analyzing the observations from 2008 and 2009. The observations were combined to build a bipartite, binary network. A bipartite network consists of two trophic levels, in this case plants and pollinators. A binary network

6

Table 1. Plant species in the networks at Ka‘ena Point, O‘ahu, Hawai‘i.

Family Species Statusa Present in 2009 Quantitative web Asteraceae Melanthera integrifolia e yes Boraginaceae Heliotropium anomalum ind yes Convolvulaceae Cuscuta sandwichiana e yes Jacquemontia sandwicensis e yes Euphorbiaceae Euphorbia degeneri e no Fabaceae Sesbania tomentosa e, * yes Goodeniaceae Scaevola coriacea e, * yes Scaevola taccada ind yes Lamiaceae Vitex rotundifolia ind yes Malvaceae Sida fallax ind yes Waltheria indica ind yes Nyctaginaceae Boerhavia repens ind yes Santalaceae Santalum ellipticum e no Scrophulariaceae Myoporum sandwicense ind yes Zygophyllaceae Tribulus cistoides ind no a Status of plant species: e=endemic, ind=indigenous, *=endangered.

7 records only the presence or absence of species-pair interactions. Combining years creates a more comprehensive estimate of interactions for a community (Nielson and Bascompte 2007, Hegland et al 2010).

Pollinator species were removed from the network if they were recorded visiting flowers only once (singletons). Removing singletons has little effect on most network indices (Dormann et al 2009). Flower visitors including ants or visitors that never were observed contacting reproductive structures were removed from the network. Though ants were frequent visitors (945 of the 1,962 observed flower visits), they were removed from the analysis (see Discussion). While we only have data on flower visitation, not pollination, we will follow other studies in referring to the network as a plant-pollinator network (Jordano 1987), based on our criteria that flower visitors must visit more than once and must touch the reproductive structures on at least one of the visits (Dupont et al 2003). Similarly, we will refer to the flower visitors as (potential) pollinators based on the same criteria (Table 2).

A quantitative network was created for 2009, in which the interactions were weighted based on how often a pollinator visited a plant species. Henceforth, I will use the term “interaction” to indicate the presence of an interaction between two species, and the term “visits” to refer to the flower visitation rate, i.e., how often the pollinator was observed visiting a flower of that plant species. For each observation session the total number of visits to flowers was summed over all visitors of the given pollinator taxon and then divided by the number of flowers observed and the observation time, to give a rate of visits per flower per hour. Note that in this usage a ‘visit’ is to a flower, not the plant; a single visitor might visit several flowers on the plant, and might even make more than one visit to the same flower. These rates, for a given plant-pollinator pair, were then averaged over all observation sessions at a given observation point. Because observations were stratified by plant species, and were not proportional to flower abundance, the average visitation rate for a plant-pollinator pair at an observation point was multiplied by the number of flowers of that plant species at that point (i.e. in the 20x20m plot) to give an estimate of the frequency of that interaction (as visits to flowers) for the entire plot. This measure is similar to weight M5 of Castro-Urgal et al. (2013), in that it takes into account observation time, the 8

Table 2. Pollinator taxa in the networks at Ka‘ena Point, O‘ahu, Hawai‘i.

Order Family Species Present in 2009 Quantitative web Coleoptera Bruchidae Stator sp. no Chrysomelidae Chrysomelidae sp. yes Coccinellidae Coccinellidae sp. larvae yes Scolytidae Scolytidae sp. yes Diptera Syrphidae Toxomerus marginatus yes Tachinidae Archytas cirphis yes Chaetogaedia monticola yes Eucelatoria armigera no Tachinidae sp. yes Unknown Diptera sp. 1 yes Unknown Diptera sp. 2 yes Unknown Diptera sp. 3 yes Unknown Diptera sp. 4 no Hemiptera Cicadellidae Cicadellidae sp. no Lygaeidae Nysius sp. yes Hymenoptera Apidae Ceratina arizonensis yes Ceratina smaragdula yes Braconidae Braconidae spp. no Chalcididae Proconura migratoria MS yes Yoshimoto (Leblanc pers comm) Colletidae Hylaeus anthracinus yes Hylaeus longiceps yes Crabronidae Tachysphex sp. no Eulophidae Eulophidae sp. yes Halictidae Lasioglossum imbrex yes Megachilidae Megachile sp. yes Vespidae Delta campaniforme no esuriens Polistes aurifer yes Lepidoptera Lycaenidae Lampides boeticus yes

9 number of flowers observed, and the abundance of the flowers. These rates were then averaged across the 10 observation points to give a measure of the community-wide frequency of visits by a given pollinator to flowers of a given plant species.

Network Indices – The bipartite package (version 2.04, Dormann et al 2009) in R 3.0.2 (R Core Team 2013) was used to calculate qualitative and quantitative network indices at the network level, trophic level, and at the species level to describe plant-pollinator interactions of this community.

We will use the terms generalist/generalized and specialist/specialized with respect to the number of interactions in which a species is involved. Connectance is the realized proportions of links in the matrix (Dunne et al 2002). In interaction networks, a nested pattern is present when each species interacts with a subset of those species interacting with the more generalist species (Bascompte et al 2003). The nested pattern will be measured with 2 indices, temperature and weighted nestedness. Temperature (T) is a nestedness measure based on the departure from a perfectly nested interaction matrix, where T°=0 is maximum nestedness (Bascompte and Jordano 2013). The isocline of a perfectly nested matrix is calculated based on the number of plants and pollinators. Unexpected absences of interactions before the isocline and unexpected presences of interactions after the isocline are recorded. The unexpected absences’ and presences’ distance from the isocline is calculated and averaged into the metric temperature (Atmar and Patterson). A maximally-nested network has species arranged by decreasing number of interactions (T=0°), and a non-nested network has a checkerboard pattern (T=100°). Weighted nestedness (WNODF) is a quantitative nested measure that is more robust to different patterns of species interactions than other measures of nestedness are, and is based on overlap and decreasing number of interactions (Almeida-Neto et al 2011). Values range from 0 (not nested) to 100 (highly nested). Compartments include species that interact more among themselves than with the rest of the network (Dormann et al 2009). Interaction evenness, based on Shannon evenness, describes the distribution and uniformity of interactions between species in a community (Tylianakis et al 2007, Kaiser-Bunbury et al 2011). The values range from 0 (uneven) to 1 (uniform). Degree is the number of links, or interactions of a species (Dormann et al 2009). Generality is calculated 10 for both trophic levels and is the weighted mean number of partners per species across a trophic level, using the Shannon diversity for frequency of interactions by species (Bersier et al 2002). Network selectivity H2’, based on Shannon entropy, measures the level of selectiveness of the network (Bluthgen et al 2006). Values range from 0 (highly opportunistic) to 1 (highly selective). Species selectiveness index d’ is low when a species’ interactions are proportionate to the abundance of its partners, reflecting opportunistic interactions, and is high when a species’ interactions are not proportional to the abundances of potential partners (selective; Bluthgen et al 2006). Values range from 0 (highly opportunistic) to 1 (highly selective).

Other indices calculated in ‘networklevel’ and ‘grouplevel’ of bipartite which are not reported in results are included in Appendix A.

Data Analysis - The observed patterns for temperature in the 2008-2009 binary dataset used a t- test against 1000 null models using the Patefield algorithm in bipartite (Dormann et al 2009). The result of the t-test was signficiant if p < 0.05.

RESULTS

The binary plant-pollinator network of 2008 and 2009 (Figure 2) includes 15 native plant species (Table 1), 2 native insect species, and 26 non-native insect taxa (Table 2), and has 119 interactions. The quantitative network of 2009 (Figure 3) includes 12 plant species (Table 1), 2 native insect species, and 19 non-native insect taxa (Table 2), with 51 interactions. Because the females of the two native Hylaeus species are indistinguishable in the field, Hylaeus were grouped together as one taxon in both networks (i.e. Forup et al 2008), yielding 27 pollinators in the 2008-2009 binary dataset and 20 in the quantitative dataset. A full report of indices calculated in the bipartite package using the functions ‘networklevel’ and ‘grouplevel’ is included in Appendix A.

Binary Network – The 2008-2009 binary dataset had a connectance of 29.4% (Table 3). The temperature index (Table 4), is significantly lower than the mean temperature of the randomized

11

Figure 2. 2008-2009 Binary plant-pollinator network of Ka‘ena Point, O‘ahu, Hawai‘i. Pollinators are represented in the columns, and plants are represented in the rows. Each shaded cell represents an interaction between two species. Matrix rows and columns are sorted according to the “binmatnest” algorithm in bipartite. Species names are abbreviated. The specific epithet or most resolved name for species and taxa are in Table 1 and Table 2 for plants and pollinators respectively.

12

Figure 3. 2009 Quantitative plant-pollinator network of Ka‘ena Point, O‘ahu, Hawai‘i. Each cell is shaded from white (no interaction and no visits) to black (interaction present and visits very frequent) indicating the strength of the interaction between the plant and pollinator. Pollinators are represented in the columns, and plants are represented in the rows. Matrix rows and columns are sorted according to the weighted interaction version of the “NODF” algorithm in bipartite. The specific epithet or most resolved name for species and taxa are in Table 1 and Table 2 for plants and pollinators respectively. The legend displays the color and corresponding number of visits in the interaction.

13

Table 3. Network Metrics of the binary and quantitative networks of Ka‘ena Point, O‘ahu, Hawai‘i.

Network Parameter 2008-2009 Binary Dataset 2009 Quantitative Dataset Number of plant species (P)a 15 12 Number of pollinator taxa (A)b 27 20 Total Interactionsc 119 51 Number of Visitsd 975 138 A/Pe 1.8 1.7 Network Sizef 405 240 Connectanceg 29.4% 21.3% a The number of plant species is represented as P. b The number of pollinator taxa is represented as A. c Total interactions is the sum of all interactions in the network. d Number of visits is the total observations of pollinators visiting flowers. e A/P is the ratio of number of pollinator taxa to the number of plant species. f Network size is the number of pollinator taxa times the number of plant species (A*P). g Connectance is the realized proportions of links in the matrix (Dunne et al 2002).

14

Table 4. Network Indices for the binary and quantitative datasets of Ka‘ena Point, O‘ahu, Hawai‘i. Standard error of the means are reported for temperature, degree, and d’.

Network Index 2008-2009 Binary Dataset 2009 Quantitative Dataset Temperaturea 31.6 25.2 Mean Null Model Temperature 38.1±0.12 15.1±0.12 WNODFb --- 18.3 Mean Degree Ac 4.4±0.73 2.6±0.59 Mean Degree P 7.9±0.85 4.3±0.58 Max Degree Ad 13 9 Max Degree P 14 8 Generality Ae --- 2.4 Generality P --- 2.9 Interaction Evennessf --- 0.52 H2g --- 0.72 Mean d’Ah --- 0.57±0.04 Mean d’P --- 0.58±0.06 a Temperature measures the departure from a perfectly nested interaction matrix, where T°=0 is maximum nestedness (Bascompte and Jordano 2013). b WNODF is weighted nestedness, a quantitative nested measure that is based on overlap and decreasing number of interactions (Almeida-Neto et al 2011). c Mean Degree is the mean number of links, or interactions of a species (Dormann et al 2009). d Max Degree is the species with the maximum number of links. e Generality is the mean number of partners per species from a trophic level weighted by their marginal totals (Bersier et al 2002). f Interaction evenness, based on Shannon evenness, describes the distribution and uniformity of interactions between species in a community (Tylianakis et al 2007). g H2 is the specialization of the network which is based on Shannon entropy (Bluthgen et al 2006). h d’ measures the level of specialization of a species (Bluthgen et al 2006), based on Shannon entropy (Bluthgen et al 2006).

15 networks, indicating a nested pattern relative to null models. This network lacks compartments (Appendix A), but does have a core block of generalists where majority of the interactions occur (Figure 2).

Quantitative Network – The quantitative network’s connectance is 21.3% (Table 3). Nestedness temperature of the observed network was (Table 4) significantly higher than the mean nestedness temperature of the randomized networks, suggesting that the observed network was less nested than the null model. WNODF was 18.3, where a value of 0 means no nested pattern and 100 indicates a perfectly nested pattern. This network lacks compartments (Appendix A), but has a core block of generalists (Figure 3). Interaction evenness (IE) of the quantitative network is 0.52. The H2 value of 0.72, indicates a selective network.

Pollinators and Plants – In both datasets, native bees (Hylaeus), and non-native small bees (Lassioglossum, Ceratina), wasps (Proconura), and flies (Tachinidae) were the most frequent flower visitors both for interactions and visits (Table 5). Hymenoptera made up the largest proportion of taxa involved in the networks. Hymenoptera were involved in the most interactions and visits compared to other insect orders. After Hymenoptera, Diptera was second for most taxa, interactions, and visits. Coleoptera, Hemiptera, and Lepidoptera made up minor proportions of the taxa, interactions, and visits.

In both years, non-native insects visited 5 plant species not visited by native Hylaeus bees, and in 2009 they provided 79.45% of visits in the quantitative network. The number of partners ranged from 1 to 13 out of 15 possible plant species with a mean degree of 4.4 ± 0.73 (S.E.M.) in the 2008-2009 binary dataset. In the 2009 quantitative dataset, non-native insects visited from 1 to 9 out of 12 possible plant species and had a mean degree of 2.6 ± 0.59. This difference is due to the subset of data used for the 2009 quantitative analysis. Overall animal generality, weighted mean number of partners, was 2.4 in the 2009 quantitative dataset.

The two native visitors, Hylaeus anthracinus and H. longiceps, provided 20.55% of the flower visits in 2009, and foraged at high frequencies and on many species, including the

16

Table 5. Insect orders represented in networks of Ka‘ena Point, O‘ahu, Hawai‘i. The number of pollinator taxa or the proportion of that insect order involved in different degrees in the network are presented for the 2008-2009 binary dataset (B) and 2009 quantitative dataset (Q). The full list of pollinator taxa and insect orders are in Table 2.

No. taxaa Taxa %b Interaction %c Visit %d Insect Order B Q B Q B Q Q Hymenoptera 11 8 40.7 40 62.2 72.5 58.0 Diptera 9 7 33.3 35 22.7 15.7 34.18 Coleoptera 4 3 14.8 15 8.4 5.9 6.07 Hemiptera 2 1 7.4 5 5.0 2.0 0.11 Lepidoptera 1 1 3.7 5 1.7 3.9 1.64 a No. taxa is the number of pollinator taxa observed for the network that fall into that insect order. b Taxa % is the proportion of pollinator taxa of an insect order out of the total number of taxa observed. c Int. % is the proportion of interactions an insect order was involved in out of the total interactions observed in the network. d Visit % is the proportion of visits an insect order was involved in out of the total visits in the network.

17 endangered Scaevola coriacea and Sesbania tomentosa. Hylaeus spp. provided high proportions of visits to 4 plant species (Table 6), including 50.38% for Scaevola coriacea, 43.22% for Sesbania tomentosa, and 32.99% of visits for Scaevola taccada. Hylaeus spp. visited most often Myoporum sandwicense, Sesbania tomentosa, and Scaevola taccada (69.47%, 14.43%, and 12.52% of Hylaeus visits, respectively).

Generalized non-native pollinators (Lasioglossum imbrex, Ceratina smaragdula, and Proconura migratoria) interacted with many plant species (Figure 2, 3). These 3 pollinators visited ≥ 80% of the 15 species in the 2008-2009 binary dataset and ≥ 58% of the 12 species in the 2009 quantitative dataset. This difference is due to the subset of data used in the 2009 quantitative analysis. These three pollinators visited the plants frequently, making up 23.6% of the visits in the quantitative network (Figure 4). Two specialized pollinators, Diptera sp. 2 and Eulophidae sp., visited Myoporum sandwicense and C. sandwichiana, respectively, and accounted for 28.2% and 9.4% of the visits in the quantitative network, respectively. In the 2008- 2009 binary dataset, Myoporum sandwicense had 12 partners, but in the 2009 quantitative dataset, it had 3. Out of these 3 pollinators, Diptera sp. 2 provided 63.3% visits to Myoporum sandwicense. This interaction was the strongest (accounting for 28.2% of visits in the network) due to the high abundance of Myoporum sandwicense flowers in a plot (5,648 flowers). C. sandwichiana had 2 visitor species in the 2008-2009 binary dataset and 1 in the 2009 quantitative dataset. Eulophidae sp. provided 100% of the 2009 visits to C. sandwichiana. In 2009, the remaining 14 out of the 20 pollinators (flies, wasps, beetles, true bugs, and a butterfly) interacted with relatively few plant species, ≤ 2 out of the 12, and did so relatively infrequently, accounting for 18.3% of the visits in the network (Figure 5).

The native coastal plant species include both common (J. sandwicensis, Sida fallax, and Scaevola taccada) and endangered species (Scaevola coriacea and Sesbania tomentosa). In both datasets, most plant species interacted with several pollinators, appearing to be generalists (mean

18

Table 6. Plant species’ proportions of visits from Hylaeus spp. from the quantitative network of Ka‘ena Point, O‘ahu, Hawai‘i.

Plant % visits from Hylaeus spp % Hylaeus spp. visits (plant view) (Hylaeus view) Scaevola coriacea 50.38 0.62 Sesbania tomentosa 43.22 14.43 Scaevola taccada 32.99 12.52 Myoporum sandwicense 32.38 69.47 Jacquemontia sandwicensis 8.27 2.36 Sida fallax 1.20 0.48 Melanthera integrifolia 0.29 0.12 Vitex rotundifolia 0.00 0.00 Waltheria indica 0.00 0.00 Heliotropium anomalum 0.00 0.00 Boerhavia repens 0.00 0.00 Cuscuta sandwichiana 0.00 0.00

19

Generalized pollinator visits 5000 4500 4000

3500 3000 Pro mig 2500 2000 Cer sma No. of visits of No. 1500 Hyl spp 1000 Las imb 500

0

Sid fal Sid

Vit rotVit

Sca tac Sca

Jac san Jac

Sca cor Sca

Mel int Mel

Cus Cus san

Wal ind Wal

Helano

Boe Boe rep Ses Ses tom Myo sanMyo Figure 4. The distribution of visits (y axis) of generalized pollinators to plant species (x axis) from the quantitative network of Ka‘ena Point, O‘ahu, Hawai‘i. The specific epithet or most resolved name for species and taxa are in Table 1 and Table 2 for plants and pollinators respectively.

20

% Pollinators' visits out of total visits

30.00%

25.00%

20.00%

15.00%

10.00%

5.00%

0.00%

Taxon % visits out of total visits total of% visits out Taxon

Dipt Dipt 2 Dipt 1 Dipt 3

Arccir

Cer ari Cer

Coc lar Coc

Scol sp Scol

Pol aur Pol

Nysi sp Nysi

Eulosp

Hyl Hyl spp

Chry sp Chry

Tach spTach

Lasimb

Pro mig Pro

Cer sma Cer

Tox Tox mar

Mega sp Mega Lam boeLam Cha monCha Figure 5. The proportion of visits (y axis) by pollinator taxa (x axis) out of the total visits in the quantitative network of Ka‘ena Point, O‘ahu, Hawai‘i. The specific epithet or most resolved name for taxa are in Table 2 for pollinators.

21 plant degree 7.9 ± 0.85 in the 2008-2009 binary dataset and 4.3 ± 0.58 in the 2009 quantitative dataset). Plant generality, weighted mean of partners, for the 2009 quantitative dataset was 3.0. In the 2009 quantitative dataset, the plants with the most visits of the network included Myoporum sandwicense (44.1% visits), Melanthera integrifolia (8.7% visits), and Sida fallax (8.2% visits; Figure 6). Results that involve Scaevola coriacea should be interpreted cautiously because only one individual grows at Ka‘ena Point.

Species selectiveness – To assess species’ selectivity, the d’ index, was evaluated in the quantitative network. Most plants and pollinators (42% and 50%, respectively) were categorized as selective (Figure 7; using the same 4 categories from Castro-Urgal and Traveset 2013).

DISCUSSION

In a Hawaiian coastal flowering-plant community consisting entirely of native plants, the pollinator community is dominated by introduced insect species. This is an extreme example of the preponderance of novel interactions between natives and non-natives within a community. The non-native pollinators comprise 95% (19/20 in the 2009 quantitative dataset) to 96% of the taxa (26/27 from the 2008-2009 binary dataset) and are involved in 79.45% of flower visits in the quantitative dataset for 2009. There are no collections of coastal pollinators in Hawaiʽi to compare this work to and the species composition of the original native pollinator community in the coastal habitat is unknown (Howarth 1985, Leblanc pers comm). Several studies have documented incorporation of non-native plants (Bjerkens et al 2007, Bartomeus et al 2008, Traveset and Richardson 2006) and/or pollinators into a plant-pollinator network (Kaiser- Bunbury and Muller 2009), but none have reported such an extreme case of a native plant community interacting primarily with non-native pollinators. A large shift in the pollinator assembly from natives to non-natives has been reported on the Ogasawara (Bonin) Islands, Mauritius, and the Galapagos Islands (Abe et al 2006, Kaiser-Bunbury and Muller 2009, Traveset et al 2013). Many native and endemic small bee taxa of the Ogasawara Islands have decreased due to predation by introduced anoles (Abe et al 2006, 2007). Non-native taxa made

22

% Plants' visits out of total visits

50.00% 45.00% 40.00% 35.00% 30.00% 25.00% 20.00% 15.00% 10.00% 5.00%

0.00%

Species % visits out of total visits total of% visits out Species

Sid fal Sid

Vit rotVit

Sca tac Sca

Jac san Jac

Sca cor Sca

Mel int Mel

Cus Cus san

Wal ind Wal

Helano

Boe Boe rep Ses Ses tom Myo sanMyo Figure 6. The proportion of visits (y axis) by plant species (x axis) out of the total visits of the quantitative network of Ka‘ena Point, O‘ahu, Hawai‘i. The specific epithet or most resolved name for species Table 1 for plants.

23

Figure 7. Categories of plant species or pollinator taxa d’ values from the quantitative network of Ka‘ena Point, O‘ahu, Hawai‘i. Species selectivity d’ measures the level of selectivity of a species (Bluthgen et al 2006). The index decreases when a species’ interactions are proportionate to abundant partners, which is considered opportunistic, and the index increases when a species’ interactions are disproportionately higher with a rare species which is considered selective. d’ categories (Castro-Urgal and Traveset 2014) are: 0 ≤ d ≤ 0.25 (Highly opportunistic), 0.25 < d ≤ 0.50 (Opportunistic) , 0.50 < d ≤ 0.75 (Selective) , 0.75 < d ≤ 1.00 (Highly selective).

24 up at least 35 out of 161 observed pollinator taxa (22%) and made 63% of the flower visits in heath vegetation in Mauritius (Kaiser-Bunbury and Muller 2009). In the Galapagos Islands dry zone habitat, alien pollinators make up 21% of the pollinator taxa and provide 38% of interactions (links). Other studies have documented interactions involving individual non-native pollinators. Pollination of native plants by non-native pollinators may be either less efficient or more efficient than pollination by native pollinators (Ings et al 2005, Kenta et al 2007, Chamberlain and Schlising 2008).

Similar to our findings, non-native species have been found to dominate various other arthropod assemblages in Hawai‘i, including lowland insects (Asquith and Messing 1993), aquatic arthropods (Englund 2002), and parasitoids of native moths feeding on native plants on Kaua‘i (Henneman and Memmott 2001). Additionally, other systems in Hawai‘i and on other islands include novel interactions between native and non-native species, which in some cases appear to maintain ecosystem functions. In New Zealand, non-native rats (Rattus rattus) and the recent colonist silvereye (Zosterops lateralis) pollinate some native forest plants (Pattemore and Wilcove 2012). On Hawai‘i Island, native and non-native pollinators shared three native plants (Clermontia parviflora, C. montis-loa, and C. hawaiiensis), but the visits were dominated by the non-native Japanese white-eye (Zosterops japonicus) (Aslan et al 2014). Z. japonicus was unable to pollinate C. hawaiiensis effectively due to a morphological mismatch, but pollinated C. parviflora and C. montis-loa. On Maui, in dry forest dominated by native plants but lacking native birds, the avian seed disperser community consists entirely of non-native species which disperse the seeds of most of the fleshy-fruited native and non-native plants (Chimera and Drake 2010). Non-native species have the potential to provide mutualistic services in the absence of native animals, though the quality and quantity of the services may not be exactly the same as those provided by the natives (Aslan et al 2012).

In some plant-pollinator communities, native generalist species could provide a way for non-native species to enter the community (Bjerkens et al 2007, Lopezaraiza-Mikel et al 2007). Many insects can easily visit the flowers of many coastal plants of Hawai‘i, including most of those present at Kaʽena Point, because the flowers are open and thus easily accessible, and are 25 available throughout much of the year. The results from our study support this idea that most plants interact with several partners (high mean plant degree, Table 4). Once involved in the network, non-native species could replace native species (Olesen et al 2002, Bartomeus et al 2008, Padron et al 2009), by usurping and disrupting the current interactions and visits in the network (Aizen et al 2008, Traveset et al 2013). Alien plants have been reported to increase the visits received by other plants by attracting more pollinators to them (Lopezaraiza-Mikel et al 2007) and, conversely, by dominating the interactions with pollinators and thus reducing visits to other plants (Bartomeus et al 2008). Overall, integration of a non-native species into a network appears to depend on the species involved and the circumstances (Vila et al 2009).

Network metrics – Both the binary and quantitative network parameters are similar to those of other small networks on islands (Olesen et al 2002, Padron et al 2009, Trojelsgaard and Olesen 2013). Compared to the average island network (Trojelsgaard and Olesen 2013), our networks have fewer species and higher connectance, which could be due to removing singletons. When compared to other island networks with relatively small sizes and in dry ecosystems, however, the total number of taxa, connectance, ratio of animals to plants, and average degree of pollinators and of plants, are broadly similar (Dupont et al 2003, Philipp et al 2006, Bartomeus et al 2008, Padron et al 2009).

Nested patterns – The Ka‘ena Point 2008-2009 binary dataset was nested based on the temperature index and Patefield null model. The 2009 quantitative dataset has a similar WNODF value as other networks on islands (WNODF ranges from 9.21 through 21.73; Traveset et al 2013, Castro-Urgal and Traveset 2014), where number of plant species ranges from 18 to 25 and number of pollinator species ranges from 60 to 93. WNODF was not tested against null models because current quantitative null models are based on discrete number of visits and are inappropriate for our flower visitation rate (Ulrich et al 2009, Gotelli and Ulrich 2011). Other studies have found mixed results when using different nestedness measures (Kratochwil et al 2009, Ulrich et al 2009), but most plant-pollinator networks are nested (Bascompte et al 2003). Our networks have a core block of generalized plants interacting with generalized pollinators (Bascompte et al 2003). Some studies report that the presence of non-natives increases 26 nestedness (Bartomeus et al 2008, Traveset et al 2013), but this likely depends on the specific system.

An unusual case in our network is the interaction between native C. sandwichiana and a non-native wasp, Eulophidae sp. C. sandwichiana does not have any other partners, and Eulophidae sp. interacts with C. sandwichiana, J. sandwicensis, and W. indica. This appears to be a native specialist interacting with a non-native specialist. However, C. sandwichiana may have other pollinators at other seasons, and other species of Cuscuta have a variety of pollinators (Wright et al 2012).

Quantified Indices – The interaction evenness (IE) of the quantitative network was lower than that of other island networks (Kaiser-Bunbury et al 2009, Traveset et al 2013, Castro- Urgal and Traveset 2014). IE declined in more heavily invaded sites in Mauritius (Kaiser- Bunbury et al 2011). Our network has mostly non-native pollinators, and the IE is similar to that of those invaded sites.

Few plant-pollinator network studies report H2 and d’ values, but the values I observed in this network fall in the upper range of values for other island pollinator networks for which these indices have been calculated (Traveset et al 2013, Castro-Urgal and Traveset 2014). The observed values indicate a selective network (Bluthgen et al 2006).

Species Interactions – Non-native, flower-visiting taxa in this study were mostly bees, wasps, and flies, and they also provided most of the flower visits. The most generalized pollinators (Ceratina smaragdula, Lasioglossum imbrex, and Proconura migratoria) interacted with the same plants, had similar degree, and provided many visits overall in the network, as did the native Hylaeus spp. However, the three non-native species distributed their visits among plant species differently than Hylaeus spp. These 4 pollinators may dominate in visits and share plant species as partners, but different plant species received varying number of visits from these frequent pollinators

27

The only native pollinator taxon was Hylaeus spp. The two native Hylaeus species are important pollinators because of the high frequency of their visits to flowers. The Hylaeus taxon represented 1/27th of the pollinator taxa in the 2008-2009 binary dataset and 1/20th of the pollinator taxa in the 2009 quantitative dataset, but interacted with many partners (10 out of 15 species in the binary network and 7 out of 12 plant species in the quantitative network) and made many visits (20.55% of the total visit frequency in the 2009 quantitative dataset). Hylaeus are recognized across the Hawaiian Islands as potentially important pollinators for native plants (Hopper 2002, Magnacca 2007, Koch and Sahli 2013, Krushelnycky 2014). Since this study was conducted in 2008 and 2009, Hylaeus populations appear to have declined throughout Hawai‘i (Magnacca and King 2013). Hylaeus bees have rarely been observed at Ka‘ena Point since 2012 (Magnacca and King 2013, Magnacca pers comm, Shay pers obs).

Another interesting change since 2008-2009 is the presence of Apis mellifera, which was not observed in 2008 or 2009 probably because of the varroa mite (Taylor pers comm). A. mellifera was quite noticeable in previous years (Taylor per obs) and visiting many flowers in spring 2014 (Shay pers obs). Studies show A. mellifera integrates into other island and native networks and dominates partners and visits (Butz Huryn 1997). A. mellifera dominated the pollinator abundance and performed 43.8% of the visits in Mauritius (Kaiser-Bunbury 2011), including 95% of the visits to an endemic shrub Bertiera zaluzania (Kaiser-Bunbury and Muller 2009). Honey bees visited 50% of the plant species in Canary Islands (Dupont et al 2003). With the apparent decline in the remaining native pollinators and return of A. mellifera, Ka‘ena Point may be approaching a completely native plant community interacting with a completely non- native pollinator community.

Some of the most active non-native flower-visiting insects in Hawai‘i’s coastal ecosystem are ants, Lasioglossum imbrex, Ceratina smaragdula, and A. mellifera (Hopper 2002, Elmore 2008, Shay Chapter 2, Shay per obs). In several other studies in this habitat, these insects have been observed as frequent flower visitors and, except for ants, as potential pollinators to Scaevola taccada, Scaevola coriacea, (Elmore 2008), Sesbania tomentosa,

28

(Hopper 2002), Sida fallax, and J. sandwicensis (unpublished data, Shay Chapter 2) These results are consistent with those of the present study at Kaʽena Points.

Not all flower visitors are effective pollinators (Popic et al 2013). One in particular that I believe was rarely legitimate pollinator to Sesbania tomentosa was Lampides boeticus. Lampides boeticus visited both V. rotundifolia and Sesbamia tomentosa, but contacted reproductive structures of only V. rotundifolia. I suspect it oviposits on Sesbania tomentosa, as suggested by previous authors (Hopper 2002, Pratt et al 2011).

Ants are an important component of this community, but one whose effects I was unable to determine. Ants visited all plant species and were recorded 945 times out of the original total of 1,962 observations of flower visitors across both years. Ants were removed from the analysis because they are only rarely pollinators (Beattie et al 1984). However, other insect interactions in the community could be affected by ants. Ants could be removing floral resources or physically interfering with pollinators on flowers (Lach 2008, Junker et al 2011, LeVan et al 2014), and could decrease pollinator populations in the habitat (Cole et al 1992, Krushelnycky and Gillespie 2008, Junker et al 2011).

CONCLUSION

I described and analyzed plant-pollinator networks to develop a basic understanding of coastal plant-pollinator interactions of Hawai‘i and of the dominating presence of non-native pollinators. I expected the network parameters and indices to be similar to those of other insular ecosystems, and in spite of the dominating presence of non-native pollinators, the network was still similar to other insular ecosystems. Most network parameters and binary indices were quite similar to values observed in other island networks (Traveset et al 2013, Trojelsgaard and Olesen 2013, Castro-Urgal and Traveset 2014).

This description of plant-pollinator interactions can assist efforts to restore and conserve the native coastal ecosystem of Hawai‘i (Medeiros et al 2013, Young et al 2013). Restoration plans should consider assessing the status of remaining native pollinators, such as Hylaeus spp.,

29 and non-native pollinators, because both are involved in many interactions and visits in the community (Magnacca et al 2007, Medeiros et al 2013). This study and other examples (Aslan et al 2012) describe how non-native species form novel interactions with native species, and can maintain an ecosystem’s function after losing most of the original native species.

30

CHAPTER II. THE POLLINATION BIOLOGY OF Jacquemontia sandwicensis (CONVOLVULACEAE)

ABSTRACT

Native plants of Hawaiʽi have received little attention regarding their basic ecology and pollination biology. As human impacts on island systems increase, so, too does the importance of understanding native species’ life histories and population dynamics for plant conservation. The breeding system and pollinator observations were assessed for the native coastal vine Jacquemontia sandwicensis (Convolvulaceae), a common plant in the coastal strand habitat. To evaluate the breeding system of this plant, field and lab experiments involving natural and hand pollination treatments were performed, as well as pollinator observations in 3 coastal sites of the southeast coast of Oʽahu. J. sandwicensis is a hermaphroditic species with a flexible, mixed mating system. High fruit set, seed set, and germinating seeds depend on pollen being deposited on the stigma by an active pollinator rather than a passive vector (wind or autogamy). However, this species still produces some seeds even in the absence of manipulation, suggesting it can reproduce when pollinators are absent or in low abundances. In the observed sites, this plant was visited by a variety of mostly non-native Hymenoptera (A. mellifera and Lasioglossum spp.). Native Hylaeus anthracinus visited J. sandwicensis, but at much lower frequency than the non- native pollinators. Non-native pollinators appear to be effective pollinators for J. sandwicensis and could provide pollination services in the face of absent or declining native pollinators.

INTRODUCTION

Native plants on islands are threatened by habitat fragmentation and loss, decline and extinction of mutualists, and invasion by non-native species (Cuddihy and Stone 1990, Whittaker and Fernandez-Palacios 2007, Kaiser-Bunbury et al 2010b), among other factors. As human impacts on island systems increase, so, too does the importance of understanding native species’ life histories and population dynamics for plant conservation (Anderson et al 2001, McMullen 2009, Caujape-Castells et al 2010). Unfortunately, basic ecological studies do not exist for many island species, common or rare (Baker 1967, Anderson et al 2001, Caujape-Castells et al 2010).

Relative to continental floras, island floras have higher proportions of plants that are dioecious, or self-compatible (Baker 1955, Carlquist 1974, Sakai et al 1995, Barrett 1996,

31

Anderson et al 2001, Schlessman et al 2014), and tend to have generalized pollinator interactions (Kaiser-Bunbury et al 2010b). Dioecious plants bear male or female flowers, forcing individuals to outcross for reproduction (Bawa 1980). Self-compatible plants are able to use their own pollen to fertilize ovules (Willmer 2011) and may or may not rely on pollinators for reproduction (Baker 1955). Dioecy and self-compatibility are contrasting strategies for colonizing an island and adapting to conditions of an island (Baker 1955, Carlquist 1974, Barrett 1996, Cheptou 2012), but both are observed in islands, including Hawaiʽi and others (McMullen 1987, Sakai et al 1995, Anderson 2001, Bernardello et al 2001, Schlessman 2014). If a plant that relies on pollinators colonizes an island, then it must find effective pollinators on the island (Kaiser- Bunbury et al 2010b). The pollinators on the island are likely a different assemblage than the plant’s origins (MacArthur and Wilson 1967). Island plants typically interact with many partners, and are considered generalists; some are even supergeneralists, having wide pollination niches and interacting with a disproportionately large fraction of the pollinator community (Olesen 2002).

Native insect pollinators and their interactions are not well known in Hawaiʽi. The pollinator community is almost certainly naturally depauperate and disharmonic compared to continental ecosystems (MacArthur and Wilson 1967, Whittaker and Fernandez-Palacios 2007). In Hawaiʽi, native pollinators could include 955 moth species (Ziegler 2002), 63 Hylaeus bees (Magnacca and King 2013), 94 Nesodynerus or potter wasps (Carpenter 2008), and 2 species of butterflies (Zimmerman 1958). The yellow-faced bees of Hawaiʽi, Hylaeus, have radiated from one colonist to over 60 species (Magnacca and Danforth 2006). Hylaeus have been observed visiting flowers of native Hawaiian plants in many habitats across the islands (Wilson et al 2010, Koch and Sahli 2013, Krushelnycky 2014), and are potential and sometimes dominant pollinators for native coastal plants, including J. sandwicensis (Hopper 2002, Chapter 1). Hylaeus and other lowland insects have been devastated by many factors (Medeiros et al 2013), including non-native insects disturbing native insects in various ways (Howarth 1985, Cole et al 1992), novel predators (i.e. Kraus et al 2012), and land use (Asquith and Messing 1993). Consequently, insects, such as Hylaeus, are declining in abundance and becoming endangered or extinct (Magnacca and King 2013). The loss of one species in a plant-pollinator relationship may reduce the fitness of its partner (Kiers et al 2010, Potts et al 2010).

32

In contrast, non-native insects are quite prevalent in lowland ecosystems. Over 2,500 species of non-native insects have established in Hawaiʽi (Pimental et al 2005). The non-native flower visitors range from likely pollinators to nectar thieves (Hopper 2002, Elmore 2008, Pleasants and Wendel 2010, Shay unpub). Non-native insects may increase or decrease the pollination activities of other insects in the community (Ings et al 2005, Kenta et al 2007, Bartomeus et al 2008).

Few studies have investigated Hawaiʽi’s coastal plants’ breeding systems and pollinators thoroughly. Some of Hawaiʽi’s coastal plants are self-compatible (Hopper 2002, Yorkston and Daehler 2006, Pleasants and Wendel 2010), hermaphroditic (Wagner et al 1990), and interact with multiple pollinators (Hopper 2002, Elmore 2008, Chapter 1).

Jacquemontia sandwicensis (pāʻūʻohiʻiaka, Convolvulaceae), is an endemic coastal morning glory found on all of the main Hawaiian islands and is the only species in its genus native to Hawaiʽi (Robertson 1974, Namoff et al 2010). J. sandwicensis is recognized as a main component and sometimes a dominant plant that defines the coastal vegetation (Richmond and Mueller-Dombois 1972, Gagne and Cuddihy 1990). In the coastal habitat, it can be found growing as a vine close to the ground, often with Sida fallax (pers obs). Despite its apparent ecological importance, little is known about its ecology or reproductive biology. Other studies that investigated Hawaiian plants mating systems have suggested J. sandwicensis is a hermaphroditic self-compatible plant based on herbarium and field samples (Carlquist 1966) and the Manual of the Flowering Plants of Hawaiʽi (Wagner et al 1990, Sakai et al 1995). However, these studies did not test the breeding systems experimentally, did not observe pollinators, and did not link the role of pollinators with J. sandwicensis’ breeding system.

In the native-dominated vegetation at Kaʽena Point in 2008 and 2009, Jacquemontia sandwicensis was observed as an important generalized plant species in terms of plant-pollinator interactions, interacting with Hylaeus spp. and 10 non-native taxa. In 2009, it received 5.9% of the community’s pollinator visits and 8.3% of those visits involved Hylaeus (Chapter 1). The objectives of this study were to 1) investigate the pollinators elsewhere on Oʽahu, in more disturbed communities, to see if its pollinators differed and 2) investigate the breeding system of J. sandwicensis to determine its reliance on pollinators for reproduction. I predicted that Hymenoptera species, including Hylaeus, if present, visit J. sandwicensis (Chapter 1).

33

Jacquemontia sandwicensis is expected to have a mixed mating system, which means a hermaphroditic plant reproduces by self and cross fertilization (Goodwillie 2005), similar to J. reclinata, a related coastal species in Florida (Pinto-Torres and Koptur 2009).

METHODS

Study sites – This study involved 4 sites along a 3.5 km stretch of coastline on the southeast of Oʽahu (Figure 1). The sites were near the Halona Blowhole (21°16’50” N, 157°40’44” W; 23.8 m elevation; BH), the Kaʽiwi Scenic Shoreline (21°17’50” N, 157°39’20” W; 3 m elevation; KI), a cliff near the Makapuʽu Lighthouse Trail (21°17’59” N, 157°39’06” W, 66 m elevation; MC), and Wawamalu Beach Park (21°17’13.98”N, 157°40’08.23”W; 2 m elevation; WP). These low elevation, low rainfall (670 mm/yr; Giambelluca et al 2013) sites experience harsh environmental conditions such as high temperature and intense solar radiation and support dry coastal plant communities with a mix of native and non-native species (Richmond and Mueller- Dombois 1972, Gagne and Cuddihy 1990).

Floral description - The inflorescence of J. sandwicensis ranges from 1 to a few flowers in a cyme in the leaf axil (Wagner et al 1999). Flower traits were measured in 22 flowers, selected from 22 different individuals from the cultivated plants used in the breeding experiment (Table 1). Basic flower morphology was measured using calipers (Raimundez-Urrutia et al 2008). The flowers vary in color from white to pale blue (Figure 2). The corolla is sympetalous and broadly campanulate, and can have 2 tall stamens and 3 short stamens. The tall stamens are of similar length to the style. Pollen-ovule ratio was based on pollen counts from a single anther, from flowers of 7 individuals, made under a dissecting microscope (Cruden 1977), and the number of ovules is always 4 (Wagner et al 1999). Mean pollen grains per anther was 587±33.25 (n = 7), and the pollen-ovule ratio was 733.8.

Breeding experiments - Cuttings of J. sandwicensis were collected at the 4 study sites (20 to 25 individuals per site) in May 2013. The cuttings were trimmed to 10 cm to 15 cm with 3 to 5 nodes, and leaves under the 3rd or 4th node from the apex were removed to reduce water loss

34

Figure 1. Map of Study sites. The upper left inset is a map of the island Oʽahu with the area highlighted with the yellow rectangle expanded into the central image. The central image has the 4 study sites, Halona Blowhole, Wawamalu Beach Park, Kaʽiwi Shoreline, and the Lighthouse Trail at Makapuʽu marked.

35

Table 1. Floral morphology of Jacquemontia sandwicensis. All traits were measured on a single flower from 22 individuals, reported in (mean in mm ± standard error). Corolla length is the length of the corolla from base to top. Corolla diameter is the diameter of the corolla at the top of the flower. Corolla base is the diameter of the corolla at the base of the flower. Style length is the length of the style. Short anther is the length of one of the 3 shorter anthers. Long anther is the length of one of 2 longer anthers. Corolla Corolla Corolla Style Short Long Length Diameter Base Length Anther Anther 13.8±0.35 13.1±0.38 5.0±0.10 8.0±0.17 4.7±0.19 7.2±0.22

36

A B

C

Figure 2. Flower of Jacquemontia sandwicensis. (A) Both male and female reproductive structures are white. The solid black arrow points to the bilobed stigma and the dashed black arrows point to the two taller stamens. (B) A male Hylaeus anthracinus visiting J. sandwicensis (Photo: Kobey Togikawa), (C) J. sandwicensis in coastal habitat at Kaʽiwi Shoreline, Oʽahu.

37 from transpiration. The cuttings were planted in 15.2 cm diameter and 12.7 cm height pots with a mix of 3 parts perlite to 1 part vermiculite, and kept well watered (Hawaiian Native Plant Propagation 2001). Once the plants grew and developed, they were transplanted into 25.4 cm diameter and 25.4 cm height pots with a mix of 4 parts commercial potting soil to 3 parts perlite to 1 part vermiculite. The plants were grown outdoors in full sunlight, where they were allowed to mature and flower. The breeding pollination treatments were conducted from 1/6/2014 through 2/11/2014.

A total of 51 plants received each of the following treatments: control (C), autogamy (A), self (S), within population (I), and between populations (O) (Table 2; Dafni et al 1992, Ushimaru and Kikuzawa 1999, Pinto-Torres and Koptur 2009). The C treatment was not bagged and the sexual structures received no manipulation; this tested for the effect of bagging (insect visitors were uncommon on the roof of the building where the plants were grown). The A treatment was bagged and the sexual structures received no manipulation. This tested if the flower facilitates self-pollination in the absence of pollinators (autogamy). In the S treatment the flower was hand- pollinated with pollen from a different flower on the same plant (geitonogamy), to test for self- incompatibility. In the I and O treatments the flower was pollinated with pollen either from a different individual from within the same site (I) or from another site (O) by hand (xenogamy or outcrossing); this assesses effects of inbreeding vs outcrossing and maximum potential seed set among populations (Dafni et al 1992, Schoen and Lloyd 1992, Pinto-Torres and Koptur 2009).

All treatments except the C used a bag made of organza fabric to isolate the flower from pollinators and pollen sources (Dafni et al 1992, Kearns and Inouye 1993). Because it is difficult to isolate individual flowers in a bag, other flowers and buds in the same axil as the treated flower were removed. Pollination was effected by using forceps to apply an open anther to a stigma (Dafni et al 1992). After treatment, flowers were bagged again, to exclude other possible pollen donors. Because flowers of J. sandwicensis last only one 1 day, the bag was removed the day following the treatment to avoid any effects of the bag on fruit development. Treated flowers were checked every other day. The developing fruit was collected when it was ripe and easily to detached. All fruit was collected by 3/14/14.

The breeding experiment was performed in the field as well on 31 plants at MC in May 2013; they received 4 treatments: C, A, F (a modified A treatment where the flower was hand

38

Table 2. Treatments in the breeding experiment of J. sandwicensis. Each treatment includes a description of the application performed, the implications of that treatment for the plant’s breeding system, and the setting where this was performed in the lab or field experiments. Treatment Application Description Tests for Lab Field Control (C) No bag. Flower accessible to Pollen limitation Yes Yes natural pollinators. Auto (A) Bagged. No hand pollination. Autogamy – self- Yes Yes Flower self-pollinates without an pollination external agent Facilitated Bagged. Pollen from the flower Self-incompatibility No Yes Selfing (F) that also receives pollen by hand pollination. Self (S) Bagged. Pollen from a different Self-incompatibility Yes No flower of the same individual by hand pollination. In (I) Bagged. Pollen from a different Seed set within a population Yes Yes individual from the same site by and inbreeding depression hand pollination. Out (O) Bagged. Pollen from a different Maximum potential seed set Yes No individual from a different site by and xenogamy or hand pollination. outcrossing

39 pollinated with its own anthers testing for self-compatibility and facilitated selfing), and I. Individuals were selected haphazardly in the site. The I treatment involved crosses with individuals that were approximately 5 m away, to reduce the likelhood of selfing between different ramets of a clone. Fruit was collected by 6/2/2014.

The C treatment was also performed in the field at 3 sites (KI, MC, and BH) on June 14, 2013, testing for pollen limitation. At each site, individuals and flowers were selected at random points. One flower on each of 40 plants at each site was tagged and left open to pollinators. Fruits were checked every other day for 25 days. They were collected when they were ripe and about to detach from the individual, or on the 25th day.

The collected fruits from the lab breeding experiment, field breeding experiment, and the control treatment at the 3 sites were stored at 20º C in wax paper envelopes in a dry environment (Pinto-Torres and Koptur 2009). Fruit and seed set were recorded for each treatment. Fruit from the field breeding experiment at MC was collected too early and was not tested for seed germination. Seeds were carefully removed from the fruit capsule and scarified with a razor blade. Each fruit’s seeds were placed on a wet paper towel in a petri dish, and placed under continuous fluorescent light (irradiance = 85-120 µmol m-2 sec-1) at 20º C. Every other day seeds were moistened and dishes rearranged to avoid positional effects. After 2 weeks, seeds with an emerged radicle or cotyledon(s) were counted as germinated (Pinto-Torres and Koptur 2009).

Pollinator Observations - I observed flower visitors to Jacquemontia sandwicensis in May and June of 2013 at 3 of the 4 study sites, BH, KI, and MC, for a total of 30 hours of observations. Ten individuals were observed per day of observations. The individuals were picked haphazardly and observed for 15 minutes each. All sites and time of day (8:00 am – 5:00 pm) were equally sampled. When a flower visitor made contact with a flower, I recorded the species of flower visitor, the number of flowers observed, the number of flowers visited by the flower visitor, the duration of the visit, if the reproductive structures of the flower were touched by the visitor, and if the visitor foraged for nectar or pollen or both. Unknown flower visitors were caught when field identification was not possible. These were later identified by Paul Krushelnycky, Sheldon Plentovich, and Karl Magnacca (pers comm). Each visitor approaching the observed flowers on

40 an individual was considered a new individual and a separate visit. The frequency of visits (FV) were calculated as: FV = (V / O) / T The number of flowers visited (V) by a flower visitor is divided by the number of flowers observed during an observation period (O), which is divided by the total number of hours observed in the field (T). T was 30 hours for calculating the FV for all 3 sites, and T was 10 hours for calculating the FV for each site. The FV is the proportion of flowers visited by flower visitors per hour (Castro-Urgal et al 2012). FVs were summed by flower visitor.

Flower abundance counts were performed at KI and MC on May 24, 2013 and BH on May 27, 2013 to help interpret overall flower visitation rates. A transect line was placed haphazardly through the middle of Jacquemontia habitat. Plots (2 m x 2 m) were laid out every 2 m on alternating sides of the transect and placed a random distance of 0 to 5 m off of the transect. Each site 30 plots were laid out and the numbers of open flowers of all flowering species were counted and recorded.

Statistics – In our study, fruit set, seeds per fruit, and germinated seeds per flower are measures of reproductive success (note that seeds germinated per flower was not tested for field breeding experiment due to collecting the fruit before it was ripe). To test for fruit set in the lab experiment, I performed the Cochran-Mantel-Haenszel test. Controlling for plant individual, the relationship between fruit production and the treatment is examined using a Cochran-Mantel- Haenszel analysis. The median number of seeds per fruit in the lab experiment and the control treatment at the 3 sites were tested using the Kruskal-Wallis one way analysis of variance on ranks. Post hoc analyses were conducted using Dunn’s Method. The median number of seeds germinated per flower was tested using the Friedman repeated measures analysis of variance on ranks. Post hoc analyses were conducted using Tukey’s procedure. In the field breeding experiment, seeds per flower were tested using one way repeated measures analysis of variance. All tests were significant if p < 0.05.

RESULTS

Breeding experiments – In the lab breeding experiment, hand-pollinated plants (S, I, and O) were about twice as likely to set fruit as unmanipulated plants, (C and A) regardless of pollen

41 source (Figure 3a). Controlling for the plant individual, a significant difference was found 2 between fruit production and the treatment, (χ 1, N = 255 = 50.1818, p < 0.0001).

The seed set per fruit of hand-pollinated plants (S, I, and O) was higher than the treatments without hand pollination in the lab breeding experiment (C and A; Figure 3b). A 2 statistical difference was found between median seed set per fruit and treatment (χ 4 = 12.016, p < 0.017), but post hoc tests did not reveal differences in pairwise multiple comparison.

Flowers from hand pollinated treatments produced about twice as many germinating seeds as the treatments without hand pollination in the lab breeding experiment (Figure 3c). 2 Median seeds germinated per flower differed between treatments (χ 4, N = 255 = 46.435, p < 0.001) (Figure 3c). Tukey tests revealed that the C and A treatments’ seeds germinated did not differ, the A and I treatments did not differ, and the hand pollinated treatments (S, I, and O) did not differ. Hand pollinated treatments differed statistically from the treatments without hand pollination.

In the field breeding experiment at MC, fruit set (Figure 4a) was higher in treatments with vectors manipulating pollen (C, F, and I) than the treatment with no manipulation available (A). However, mean seeds per flower in the field breeding experiment did not differ between treatments (F 3, 30 = 1.124, p = 0.344; Figure 4b).

The C treatments’ fruit set ranged from 54% at MC to 88% at BH (Figure 5a). The 2 median seeds per fruit did not differ between the 3 sites (χ 2=1.684, p=0.431; Figure 5b). Seeds germinated per flower were similar at BH and KI, but at MC, very few seeds germinated (Figure 5c; see Discussion).

Pollinator observations - Flower visitors to J. sandwicensis were mostly bees (96%) and wasps (4%) (Table 3). Overall, non-native pollinators, A. mellifera and Lasioglossum spp., provided most of the flower visits (37.8% and 45.6% respectively) to J. sandwicensis. Three species of Lasioglossum were present in and near the sites, L. imbrex, L. microlepoides, and L. impavidum (Magnacca, pers comm). Not all Lasioglossum individuals could be distinguished in the field, and were not collected. Other non-native pollinators included Tachysphex sp. which is the only wasp observed, and C. smaragdula. Native pollinator Hylaeus anthracinus, was present at KI and MC, and provided 4.2% of the total visits. Males of Hylaeus species can be identified

42

A 1

0.8

0.6

0.4

fruit per flower per fruit 0.2

0

B 4

3

2

seeds per fruit per seeds 1

0

C

2.5 c c

2 bc

1.5

1 ab a 0.5

0 germinated seeds per flower per seeds germinated C A S I O

Figure 3. Lab breeding experiment (N = 51 plants). Treatments are: C=Control, no pollen applied, open to pollinators; A=Autogamy, no pollen applied, pollinators excluded; S=Self, pollen applied from a different flower of the same individual, pollinators excluded; I=In, pollen applied from a different individual’s flower that was from the same site, pollinators excluded; O=Out, pollen applied from a different individual’s flower that was from a different site, pollinators excluded. Treatments with the same letter do not differ significantly from each other (P, 0.05). From top to bottom, (A) Fruit set per flower, (B) Mean seed set per fruit, and (C) Mean germinated seeds per flower and standard error. Treatments sharing the same letter (C) did not differ statistically in median seeds germinated per flower.

43

A 1

0.8

0.6

0.4

fruit per flower per fruit 0.2

0 B

2.5

2 1.5 1

0.5 seeds per flower per seeds 0 C A F I

Figure 4. Field breeding experiment (N = 31 plants). Treatments are: C=Control, no pollen applied, open to pollinators; A=Autogamy, no pollen applied, pollinators excluded; F=Facilitated selfing, pollen applied from within the flower to the same flower, pollinators excluded; I=In, pollen applied from a different individual’s flower that was from the same site, pollinators excluded. From top to bottom, (A) Fruit set per flower and (B) Mean seeds per flower and standard error.

44

A 1

0.8

0.6

0.4

fruit per flower per fruit 0.2

0 B 4

3

2

seeds per fruit per seeds 1

0 C

1.5

1

0.5

0 MC KI BH -0.5 germinated seeds per flower per seeds germinated Figure 5. Control treatment (no manipulation, open to natural pollinators) at 3 sites and results. From an initial treatment of 40 flowers, I retrieved 39 fruits at Makapuʽu Cliff (MC), 36 at Kaʽiwi Shore (KI), and 40 at Halona Blowhole (BH). From top to bottom, (A) Fruit set per flower, (B) Mean seed set per fruit, and (C) Mean germinated seeds per flower and standard error.

45

Table 3. Pollinators at field sites and their proportions of visits to J. sandwicensis. The % of visits for all 3 sites (V), Kaʽiwi Shore (KI), Halona Blowhole (BH), and Makapuʽu Cliff (MC). The % of observations out of total number of observations when a flower visitor contacted the reproductive structures (RS), collected nectar (N), or collected pollen (P). Some flower visitors collected both pollen and nectar in a visit. The duration of the visit (V Dur) is mean seconds / observation.

Pollinator Family V KI BH MC RS N P V Dur

Apis mellifera Apidae 37.8 53.3 0.0 67.4 13/15 14/15 0 8.9

Ceratina smaragdula Apidae 8.3 0.0 18.9 6.6 2/3 1/3 2/3 92.3

Hylaeus anthracinus Colletidae 1.0 0.0 0.0 4.1 1/1 0 1/1 60.0

Hylaeus Female Colletidae 3.2 4.0 0.0 6.6 2/3 1/3 1/3 26.7

Lassioglossum spp. Halictidae 45.6 37.6 81.1 7.0 21/27 19/27 15/27 41.3

Tachysphex sp. Crabronidae 4.0 5.1 0.0 8.3 2/3 3/3 0 12.8

46 because of facial markings, but females are indistinguishable in the field. Ants also visited J. sandwicensis 28 times, but were removed from further analysis because they are unlikely pollinators.

Sites varied in pollinator assemblages and in the frequencies of their visits (Table 3). The most striking pollinator community was at the BH site because two pollinators, Lasioglossum spp. and C. smaragdula visited J. sandwicensis, with Lasioglossum spp. providing 81.1% of the visits. At KI, 4 species were present, but A. mellifera provided 53.3% of the visits. Lasioglossum spp. also provided many visits, 37.6% at KI. A. mellifera provided the most visits, 67.4%, at MC.

All flower visitors were considered potential pollinators because they contacted reproductive structures often (>67%) during observations (Table 3). Lasioglossum spp., H. anthracinus, Hylaeus female sp., and C. smaragdula were all observed collecting pollen and nectar, whereas A. mellifera and Tachysphex sp. were observed collecting only nectar. Lasioglossum spp. were often covered in white pollen (assumed to be J. sandwicensis). They often crawled over anthers and contacted the stigma. They searched the base of the flower, presumably for nectar, and groomed in J. sandwicensis flowers. A. mellifera foraged at the base of the flower (for nectar presumably) and accidentally contacted anthers or the stigma. Lasioglossum spp. spent more time on a flower (mean 41.3 s/obs) than A. mellifera (mean 8.9 s/obs). Activity was throughout the day, approximately 8 AM through 4 PM

Flower abundance of J. sandwicensis and other species varied by site (Table 4). The abundance of Jacquemontia flowers per meter squared was higher at KI and MC (2.13 and 2.32 respectively) than at BH (0.46). KI had the highest species richness (6 native species, 6 non- native species), followed by BH (4 native species, 6 non-native species), and MC (4 native species, 4 non-native species).

DISCUSSION

Breeding system – The breeding experiments demonstrated that J. sandwicensis has a mixed mating system, reproducing equally well through both selfing and outcrossing. High levels of fruit set, seed set per fruit, and seeds germinating per flower depended on pollen getting deposited on the stigma by a pollinator. In the absence of a pollen vector, fewer fruits and seeds

47

Table 4. Summary of flowering species at sites. This describes the mean flowers available per meter squared and flowering plant species richness per site. MC = Makapuʽu Cliff Lighthouse Trail, KI = Kaʽiwi Shoreline, BH = Halona Blowhole, N=native, and A=alien. Plant species Origins MC KI BH Boerhavia repens N 1.28 4.57 0.00 Jacquemontia sandwicensis N 2.32 2.13 0.46 Myoporum sanwicense N 0.00 1.86 0.00 Scaevola taccada N 0.00 0.50 0.00 Sida fallax N 2.05 1.76 0.24 Waltheria indica N 0.18 0.12 0.18 Asystasia gangetica A 0.00 0.00 0.10 Emilia spp. A 0.00 0.34 0.15 Euphorbia hirta A 0.28 0.25 0.02 Fabaceae sp. A 0.00 0.08 0.00 Lantana camara A 0.00 0.00 0.02 Leucaena leucocephala A 0.02 0.06 0.00 Sonchus oleraceus A 0.00 0.91 0.61 Stapelia gigantea A 0.03 0.00 0.00 Tridax procumbens A 0.18 0.00 0.04 Species richness 6N, 9A 4N, 4A 6N, 6A 4N, 6A

48 were produced, indicating that while they can self, the frequency of this in the field is probably low without pollinators. Reproductive success was independent of pollen source, suggesting that facilitated selfing, geitonogamy, and xenogamy are all successful modes of reproduction (Lloyd 1992). High levels of fruit, seeds, and seeds germinating from S and F treatments in the lab and field, respectively, demonstrate this species is self-compatible (Dafni et al 1992, Lloyd and Schoen 1992). High levels of germination from within-population pollen donors suggest that inbreeding is not limiting production of viable seeds, but a better evaluation would be growing the progeny past germination (Charlesworth and Charlesworth 1987, Goodwillie 2005). In the wild, J. sandwicensis needs a pollinator to handle and manipulate pollen onto a stigma, regardless of the pollen source, for high reproductive output.

Mixed mating systems are the most prevalent in animal-pollinated plants and considered the “best of both worlds” because they are able to outcross and self-pollinate (Lloyd 1992, Goodwillie 2005). Autogamy could provide reproductive assurance when mates or pollinators are scarce (Kalisz and Vogler 2003, Brys and Jacquemyn 2011), but could have costs due to gamete and pollen discounting (Lloyd 1992) and inbreeding depression (Charlesworth and Charlesworth 1987). Other modes of selfing that rely on pollinators, facilitated selfing and geitonogamy, do not provide reproductive assurance (Kalisz and Vogler 2003) and may be unavoidable by-products of selection for outcrossing success if multiple flowers are required to attract a pollinator (Barrett 2003, Goodwillie 2005).

Although autogamy is possible in J. sandwicensis, it is certainly not as effective as pollen getting deposited on the stigma by a pollinator (Lloyd 1992). Placement of sexual organs in a flower can affect the mating systems in a plant, and J. sandwicensis typically has 2 different length anthers in every flower. Similar length anthers and style could allow a pollinator to precisely place the pollen on the stigma, but could easily lead to facilitated selfing, where different length anther filament from the style in a flower could help prevent facilitated selfing, but would also result in less precise pollen placement (Barrett 2002; Figure 2). J. sandwicensis flowers typically have 3 short anthers and 2 long anthers, which could allow for outcrossing and selfing. I did not emasculate any flowers and did not test for all types of autogamy (Schoen and Lloyd 1992), but I suspect that wind is a possible mechanism which causes the anthers to brush against the stigma lobes, transferring pollen in the process. The breeding experiment showed the

49

A treatment does not produce fruits as readily, as many seeds, or as many seeds that will germinate compared to treatments with pollen being transferred by hand. The breeding system of J. sandwicensis favors pollinators depositing pollen, but some seed set is possible without pollinators in the community.

The field breeding experiment and C from MC, BH, and KI show that pollen limitation did not drastically impact fruit set, seed set, or germinating seeds. The 3 sites produced fruit, seeds, and germinating seeds. The number of germinating seeds at BH and KI and at MC from the field breeding experiment fell within the range of germinating seeds from the treatments of breeding experiment. This could suggest there is some pollen limitation in the wild and could select selfing mechanisms for reproductive assurance (Goodwillie 2005). However, pollen limitation is reported widely (Burd 1994). Hand pollination and C treatments in the field produced higher seed set compared to the C and A treatments in the lab and A treatments in the field breeding experiments, which were not exposed to pollinators or hand pollination. While MC had lower fruit set and seed set per fruit than the other sites, it had extremely low seeds germinated per flower. There were fewer pollinator visits overall in this site, but this is still lower than lab and field treatments that received no pollen manipulation. Even if there were no pollinators in this site, it should have produced a similar number of seeds germinated per flower as the C and A lab and A field treatments. I suspect an environmental factor (i.e. heat or drought) interfered with reproduction at this site (Finkelstein et al 2002, Liu et al 2004, Su et al 2013).

The breeding system of J. sandwicensis contrasts with that of the related, endangered species from the Florida coast, J. reclinata (Pinto-Torres and Koptur 2009). In J. reclinata, nearby individuals were assumed to be more closely related than individuals further away (Maschinski et al 2006). In J. reclinata, reproductive success, defined as fruit set per flower, seed set per flower, and seed set per fruit, generally increased as the degree of outcrossing (distance from pollen donor) increased (Pinto-Torres and Koptur 2009). However, J. reclinata is capable of producing some seeds via autogamy and geitonogamy. For J. sandwicensis, deposition of pollen by an active vector seems to be the most important factor, and plant relatedness has negligible effect on production of a first generation of viable seeds. The breeding system has been evaluated for many species of Convolvulaceae. Species of Calystegia

50

(Ushimaru and Kikuzawa 1999), Convolvulus (Suarez et al 2004), Ipomoea (Devall and Thein 1992), and Merremia (Raimundez-Urrutia et al 2008) vary in self-compatibility, autogamy, and outcrossing.

Flower visit observations - J. sandwicensis blooms throughout the year (Wagner et al 1999), and its flowers start to open in the morning around 7 a.m. and close in early-to-late afternoon, with most closed around 5 p.m. (pers obs). Across the 3 sites, I observed 5 potential pollinator genera (4 bee and 1 wasp) active throughout the day (approximately 8 a.m. through 4 p.m.). These potential pollinators all frequently contacted reproductive structures. The majority of visits (83.4%) involved non-native bees, A. mellifera and Lasioglossum spp. One native bee, H. anthracinus and presumably females of H. anthracinus, are potential pollinators, but had very few visits in this study.

Although the three field sites were separated by < 3.5 km, pollinator abundance and activity differed. At KI, I observed 4 flower visitors (A. mellifera and Lasioglossum spp. providing many visits), frequent visits, and high numbers of flowers in the site. At MC, J. sandwicensis had high numbers of flowers through the site, and 6 flower visitors were observed (A. mellifera provided the most visits). Altogether these 6 flower visitors provided about half the visitation observed at KI or BH. Only 2 flower visitors were observed frequently visiting flowers at BH (Lasioglossum spp. dominating the visits). However, there were far fewer flowers in this site than the others. This site also contained more non-native plants than the other two sites.

Non-native species are potentially providing most pollination services for J. sandwicensis at these 3 sites because of their dominating visitation rates. Hylaeus species across Hawaiʽi are declining in abundance (Magnacca and King 2013), and they were uncommon visitors in this study. However, they were frequent and potential pollinators to another native coastal plant, Sida fallax, in these same sites and time period (Shay unpub). Potentially, Lasioglossum spp. and A. mellifera could become important substitute pollinators for J. sandwicensis owing to their frequent visits during which they contact reproductive structures. Additionally, these non- natives were the dominant visitors at BH and KI. The C treatment from BH and KI both had higher fruit set, seeds per flower, and seeds germinated per flower than the C and A lab

51 treatments. Assuming that most of the reproductive success is due to the dominant visitors, Lasioglossum spp. and A. mellifera, they appear to be satisfactory pollinators for J. sandwicensis.

Carpenter bees, Xylocopa sonorina, were not recorded visiting J. sandwicensis flowers, but were occasionally observed visiting other individuals of J. sandwicensis at KI. X. sonorina was the only pollinator in the community that appeared to experience difficulty visiting J. sandwicensis due to its large size preventing it from reaching the base of the flower for nectar. Jacquemontia sandwicensis flower could not support the carpenter bees visiting and would bend over if they visited.

The native yellow-faced bees, Hylaeus anthracinus and H. longiceps, and a variety of non-native bees, wasps, and flies were observed visiting Jacquemontia sandwicensis at Kaʽena Point, Oʽahu in 2008 and 2009 (Chapter 1). Lasioglossum imbrex, Ceratina smaragdula,and Hylaeus anthracinus were the most frequent visitors. A variety of native and non-native pollinators can visit J. sandwicensis, but most visitors are bees and wasps. J. sandwicensis’ ability to interact with a variety of pollinators, both native and non-native, agrees with Baker’s Law (Baker 1955, Baker 1967) and plant-pollinator interactions of islands (Kaiser-Bunbury 2010b).

Jacquemontia sandwicensis population could be sustained by its ability to form interactions with many pollinators and through occasional autogamy when mates or pollinators are scarce (Lloyd and Schoen 1992, Goodwillie 2005). Jacquemontia sandwicensis is thought to have colonized Hawaiʽi by fruits or seeds drifting in the ocean through the Central American Seaway (Namoff et al 2010), before the Isthmus of Panama fully closed around 3 to 4 million years ago (Coates and Stallard 2013). The known native pollinators, Hylaeus spp., are estimated to have colonized Hawaiʽi 500,000-700,000 years ago (Magnacca and Danforth 2006). Prior to the arrival of Hylaeus, J. sandwicensis would have required other pollinators or had to rely on autogamy for reproduction by seed. The same conditions exist for the coastal Gossypium tomentosum (Pleasants and Wendel 2010), suggesting that at least these two coastal plant species are able to adapt to changing pollinator assemblage or can rely on their mixed mating systems to survive.

52

Other species of Jacquemontia elsewhere are visited mostly by small to medium sized bees. In Florida, J. reclinata was visited by 18 taxa of Hymenoptera (15 taxa of bees) (Pinto- Torres and Koptur 2009). In Brazil, J. montana, J. nodiflora, and J. multiflora are visited by predominantly small to medium sized bees (Silva et al 2010, Kiill and Simao-Bianchini 2011, Kiill and Ranga 2000). Other species of Convolvulaceae also report high visitation from mostly bees and the bee syndrome, melittophily, (Galetto and Bernardello 2004, Maimoni-Rodella and Yanagizawa 2007, Pick and Schlindwein 2011).

Body size in bees correlates to their foraging range (Greenleaf et al 2007), which could affect pollen flow between individuals and populations of J. sandwicensis. The observed bees in this study were both small (Lasioglossum spp., Hylaeus females) and medium-sized bees (A. mellifera). The small bees are predicted to move pollen within a population due to their smaller foraging range of 150-600 m, whereas A. mellifera could move over >1 km depending on resources in the landscape (Gathmann and Tscharntke 2002, Greenleaf et al 2007). The sites I observed are approximately 3.5 km (BH to MC), 2.9 km (BH to KI), or 586 m (KI to MC) apart, but the vegetation is semi-continuous through these sites and could allow for bee foraging across sites.

Often, ants were observed visiting J. sandwicensis flowers (28/80 observed visits). Ants were excluded from the analysis because they are unlikely pollinators (Beattie et al 1984, McMullen 2009). Ants have been observed removing floral resources preferentially at native Hawaiian plants due to lack of barriers or defenses against ants (Junker et al 2011). Pollinators could also be impacted by ants due to physical interference on the flowers (Lach 2008) and by predation and competition in the habitat (Cole et al 1992, Krushelnycky and Gillespie 2008).

CONCLUSION

J. sandwicensis, a common plant in the coastal strand habitat, has a flexible, mixed mating system. High fruit set, seed set, and germinating seeds depend on pollen being deposited on the stigma by an active pollinator rather than a passive vector (wind or autogamy). However, this species still produces some seeds even in the absence of manipulation, suggesting it can reproduce when pollinators are absent or in low abundances. In the observed sites, this plant was visited by a variety of mostly non-native Hymenoptera (A. mellifera and Lasioglossum spp.).

53

Non-native pollinators appear to be effective pollinators for J. sandwicensis and could provide pollination services in the face of absent or declining native pollinators.

54

APPENDIX A. NETWORK AND GROUP INDICES OF KAʽENA POINT

Network and group indices of pollinators (A) and plants (P) calculated using the bipartite 2.04 package (Dormann et al 2009) for the binary and quantitative datasets of Kaʽena Point not reported in Chapter 1.

Index Definition (Dormann et al 2009) Binary Quantitative Web Asymmetry Balance between numbers in the two levels: 0.29 0.25 positive values indicate more A species, negative more P species Links per species Mean number of links per species (qualitative): 2.8 1.6 sum of links divided by number of species. Number of Compartments are sub-sets which are not 1 1 compartments connected to another compartment. Mathematically, they are Jordan blocks. Cluster coefficient The cluster coefficient is the average cluster 0.20 0.08 coefficients of its members, i.e. the number of realized links divided by the number of possible links. Weighted A nestedness version that considers interaction 0.26 0.20 nestedness (WINE) frequencies. It ranges between 1 (nested) and 0 (chaotic). Interaction strength Dependence asymmetry is a measure of 0 0.16 asymmetry specialization, across both trophic levels. (dependence) Positive values indicate higher dependence in the A trophic level. Specialization Asymmetry (A vs. P) of specialization (based on -0.33 -0.12 asymmetry d’) is insensitive to the dimensions of the web. Can be weighted by abundance or d' values log transformed. Positive values indicate a higher specialization of the A trophic level.

55

Appendix A. (continued) Network and group indices of Kaʽena Point.

Linkage density Marginal totals-weighted diversity of 8.4 2.6 interactions per species (quantitative), equivalent to the average of generality of A and P. Weighted Linkage density divided by A + P. This will 0.20 0.08 connectance respond to whether non-interacting species are included or not. Shannon diversity Shannon’s diversity of interactions. 4.8 2.9 Alatalo interaction A different measure for web entry evenness. 1 0.52 evenness Mean no. shared Mean number of partners shared among A. 1.4 0.6 partners A Mean no. shared Mean number of partners shared among P. 3.7 1.6 partners P Cluster coefficient Cluster coefficient for A. 0.50 0.36 A Cluster coefficient Cluster coefficient for P. 0.34 0.20 P Weighted cluster Weighted version uses interactions as weights 0.96 0.63 coefficient A unless the data are binary. Weighted cluster Weighted version uses interactions as weights 0.75 0.34 coefficient P unless the data are binary. Niche overlap A Mean similarity in interaction pattern between 0.24 0.11 pollinators, calculated as Horn’s index. Values near 0 indicate no common use of niches, 1 indicates perfect niche overlap. Niche overlap P Mean similarity in interaction pattern between 0.44 0.24 plants, calculated as Horn’s index. Values near 0 indicate no common use of niches, 1 indicates perfect niche overlap.

56

Appendix A (continued) Network and group indices of Kaʽena Point. Togetherness A Mean number of co-occupancies across all 0.11 0.07 species-combinations; the whole matrix is scanned for submatrices of the form (0, 0, 1, 1), representing perfect matches of co-presences and co-absences. Togetherness P Mean number of co-occupancies across all 0.28 0.21 species-combinations; the whole matrix is scanned for submatrices of the form (0, 0, 1, 1), representing perfect matches of co-presences and co-absences. C score A Mean number of checkerboard combinations 0.52 0.64 across pollinators. Values close to 1 indicate that there is evidence for disaggregation. Values close to 0 indicate aggregation of pollinators. C score P Mean number of checkerboard combinations 0.28 0.39 across plants. Values close to 1 indicate that there is evidence for disaggregation. Values close to 0 indicate aggregation of plants. V ratio A Variance-ratio of species numbers to interaction 4.72 3.45 numbers within A. Values larger than 1 indicate positive aggregation, values between 0 and 1 indicate disaggregation of pollinators. V ratio P Variance-ratio of species numbers to interaction 2.54 1.51 numbers within P. Values larger than 1 indicate positive aggregation, values between 0 and 1 indicate disaggregation of plants.

57

Appendix A (continued) Network and group indices of Kaʽena Point. Discrepancy A Nestedness measure that first sorts the matrix by 38 20 marginal totals. Then, A interactions are maximally compacted, yielding P. Discrepancy is the number of disagreements of A. Discrepancy P Nestedness measure that first sorts the matrix by 40 25 marginal totals. Then, A interactions are maximally compacted, yielding P. Discrepancy is the number of disagreements of P. Extinction slope A Slope of the secondary extinction sequence in A, 3.27 1.58 following extermination of species in P. Extinction slope P Slope of the secondary extinction sequence in P, 7.32 3.67 following extermination of species in A. Robustness A Calculates the area below the secondary 0.76 0.61 extinction curve. Robustness P Calculates the area below the secondary 0.87 0.78 extinction curve. Functional Functional complementarity measures of niche 47.4 25108 complementarity A complementarity as the total branch length of a “functional dendrogram” based on qualitative differences of interactions of one level with the other. Functional Functional complementarity measures of niche 39.4 26280 complementarity P complementarity as the total branch length of a “functional dendrogram” based on qualitative differences of interactions of one level with the other. Partner diversity A (Weighted) mean Shannon diversity of the 1.8 0.66 number of interactions for the species of A. Partner diversity P (Weighted) mean Shannon diversity of the 2.2 0.97 number of interactions for the species of P.

58

APPENDIX B. 2009 QUANTITATIVE MATRIX OF KAʽENA POINT

boe

Dipt Dipt 2 Dipt 1 Dipt 3

Cer Cer ari

Arc Arc cir

Pol aur

Coc lar Coc Scol sp

Eulo sp sp Nysi

Hyl Hyl spp Tach sp Chry sp

Pro mig

Las Las imb

Cer Cer sma

Tox mar sp Mega

Lam

Cha mon Cha

7

0 0 0 0 0 0 0 0 0 0 0 0

30.6 67.2 30.4 30.4

414.4 521.6 281.6

1 104.61

Jac san Jac

7 7

0 0 0 0 0 0 0 0 0 0 0 0 0 0

6.8

560 278

579.6

816.6 100.6

Mel int Mel

0 0 0 0 0 0 0 0 0 0 0 0 0 0

944

53.33 188.8 53.33

170.18 694.06

Sca Sca tac

7

7

7

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

113

47.2

26.6

Sid fal

554.6

1482.6

5

9

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

336

179.2

Vit Vit rot

171. 144.52 385.78

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

100 150

96.43 42.86

128.57

Wal Wal ind

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

800 480

272.44 298.67

Ses tom Ses

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

52

49.33 16.73 29.66

Hel Hel ano

0.67

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

512

385 7530.67

Myo san Myo

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

264

552.89

Boe rep Boe

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

34.1

33.58

Sca Sca cor

0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

2214 Cus san Cus

59

LITERATURE CITED

Abe, T. 2006. Threatened pollination systems in native flora of the Ogasawara (Bonin) Islands. Ann. Bot. 98:317-334.

Abe, T., S. Makino, and I. Okochi. 2008. Why have endemic pollinators declined on the Ogasawara Islands? Biodivers. Conserv. 17:1465-1473.

Aizen, M.A., C.L. Morales, and J.M. Morales. 2008. Invasive mutualists erode native pollination webs. PLoS Biol. 6:396-403.

Almeida-Neto, M., and Ulrich, W. 2011 A straightforward computational approach for measuring nestedness using quantitative matrices. Environ. Modell. and Softw. 26:173–178.

Anderson, G.J., G. Bernardello, T.F. Stuessy, and D.J. Crawford. 2001. Breeding system and pollination of selected plants endemic to Juan Fernandez Islands. Am. J. Bot. 88:220-233.

Aslan, C.E., E.S. Zavaleta, D. Croll, and B. Tershy. 2012. Effects of native and non-native vertebrate mutualists on plants. Conserv. Biol. 26:778-789.

Aslan, C.E., E.S. Zavaleta, B. Tershy, D. Croll, and R.H. Robichaux. 2014. Imperfect replacement of native species by non-native species as pollinators of endemic Hawaiian plants. Conserv. Biol. 28: 478-488.

Asquith, A., and R.H. Messing. 1993. Contemporary Hawaiian insect fauna of a lowland agricultural area on Kauaʽi: implications for local and island-wide fruit fly eradication programs. Pac. Sci. 47:1-16.

Baker, H.G. 1955. Self-compatibility and establishment after ‘long-distance’ dispersal. Evolution 9:347-349.

Baker, H.G. 1967. Support for Baker’s Law – as a rule. Evolution 21:853-856.

Barrett, S.C.H., B. Emerson, and J. Mallet. 1996. The reproductive biology and genetics of island plants. Philos. T. Roy. Soc. B. 351:725-733.

Barrett, S.C.H. 2003. Mating strategies in flowering plants: the outcrossing-selfing paradigm and beyond. Philos. T. Roy. B. . Trans. R. Soc. Lond. B 358:991-1004.

Bartomeus, I., M., Vila, and L. Santamaria. 2008. Contrasting effects of invasive plants in plant- pollinator networks. Oecologia 155:761-770.

Bascompte, J., P. Jordano, C.J. Melian, and J.M. Olesen. 2003. The nested assembly of plant- animal mutualistic networks. PNAS 100:9383-9387.

60

Bascompte, J., and P. Jordano. 2007. Plant-animal mutualistic networks: The architecture of biodiversity. Annu. Rev. Ecol. Evol. Syst. 38:567-593.

Bascompte, J., and P. Jordano. 2013. Mutualistic Networks. In: Levin, S.A and H.S. Horn, editors. 2013. Monographs in Population Biology Series, no. 53. Princeton, Princeton University Press.

Bawa, K.S. 1980. Evolution of dioecy in flowering plants. Ann. Rev. Ecol. Syst. 11:15-39. Beattie, A.J., C. Turnbull, R.B. Know, and E.G. Williams. 1984. Ant inhibition of pollen function: a possible reason why ant pollination is rare. Am. J. Bot. 71:421-426.

Beattie, A.J., C. Turnbull, R.B. Know, and E.G. Williams. 1984. Ant inhibition of pollen function: a possible reason why ant pollination is rare. Am. J. Bot. 71:421-426.

Bernardello, G., G.J. Anderson, T.F. Stuessy, and D.J. Crawford. 2001. A survey of floral traits, breeding systems, floral visitors, and pollination systems of the Angiosperms of the Juan Fernandez Islands (Chile). Bot. Rev. 67:255-308.

Bersier, L.F., C. Banasek-Richter, and M.F. Cattin. 2002. Quantitative descriptors of food-web matrices. Ecology 83:2394-2407.

Bjerknes, A.L., O. Totland, S.J. Hegland, and A. Nielsen. 2007. Do alien plant invasions really affect pollination success in native plant species? Biol. Conserv. 138:1-12.

Bluthgen, N., F. Menzel, and N. Bluthgen. 2006. Measuring specialization in species interaction networks. BMC Ecol. 6:9.

Brys, R., and H. Jacquemyn. 2011. Variation in the functioning of autonomous self-pollintion, pollinator services and floral traits in three Centaurium species. Ann. Bot. 107:917-925.

Burd, M. 1994. Bateman’s principle and plant reproduction: The role of pollen limitation in fruit and seed set. Bot. Rev. 60:83-139.

Butz Huryn, V.M. 1997. Ecological impacts of introduced honey bees. Q. Rev. Biol. 72:275-297.

Carlquist, S. 1966. The biota of long-distance dispersal. IV. Genetic systems in the floras of oceanic islands. Evolution 20:433-455.

Carlquist, S. 1974. Island biology. Columbia University Press, New York.

Carpenter, J.M. 2008. Review of Hawaiian Vespidae (Hymenoptera). Bishop Museum Occasional Papers 99:1-18.

Castro-Urgal, R., C. Tur, M. Albrecht, and A. Traveset. 2012. How different link weights affect the structure of quantitative flower-visitation networks. Basic Appl. Ecol. 13:500-508.

61

Castro-Urgal, R., and A. Traveset. 2014. Differences in flower visitation networks between an oceanic and a continental island. Bot. J. Linn. Soc. 174: 478-488.

Chamberlain, S.A., and R.A. Schlising. 2008. Role of honey bees (Hymenoptera: Apidae) in the pollination biology of a California native plant, Triteleia laxa (Asparagales: Themidaceae). Environ. Entomol. 37:808-816.

Caujape-Castells, J., A. Tye, D.J. Crawford, A. Santos-Guerra, A. Sakai, K. Beaver, W. Lobin, F.B.V. Florens, M. Moura, R. Jardim, I. Gomes, and C. Kueffer. 2010. Conservation of oceanic island floras: present and future global challenges. Perspect. Plant Ecol. Evol. Syst. 12:107-129.

Charlesworth, D., and B. Charlesworth. 1987. Inbreeding depression and its evolutionary consequences. Ann. Rev. Ecol. Syst. 18:237-268.

Cheptou, P.-O. 2012. Clarifying Baker’s Law. Annals of Botany 109:633-641.

Chimera, C.G., and D.R. Drake. 2010. Patterns of seed dispersal and dispersal failure in a Hawaiian dry forest having only introduced birds. Biotropica 42:493-502.

Christianou, M., and B. Ebenman. 2005. Keystone species and vulnerable species in ecological communities: strong or weak interactions? J. Theo. Biol. 235:95-103.

Coates, A.G., and R.F. Stallard. 2013. How old is the isthmus of Panama? B. Mar. Sci. 89:801- 813.

Cole, F.R., A.C. Medeiros, L.L. Loope, and W.W. Zuehlke. 1992. Effects of the Argentine ant on arthropod fauna of Hawaiian high-elevation shrubland. Ecology 73:1313-1322.

Colwell, R.K., R.R. Dunn, and N.C. Harris. 2012. Coextinction and persistence of dependent species in a changing world. Annu. Rev. Ecol. Evol. Syst. 43:183-203.

Cox, P.A., and T. Elmqvist. 2000. Pollinator extinction in the Pacific islands. Conserv. Biol. 14:1237-1239.

Cruden, R.W. 1977. A conservative indicator of breeding systems in flowering plants. Evolution 31:32-46.

Cuddihy, L.W., and C.P. Stone. 1990. Alteration of native Hawaiian vegetation: effects of humans, their activities and introductions. University of Hawai‘i Cooperative National Park Resources Studies Unit, University of Hawai‘i Press, Honolulu, Hawai‘i.

Dafni, A., P.G. Kevan, and B.C. Husband. 2005. Practical Pollination Biology. Enviroquest, Ontario, Canada.

Devall, M.S., and L.B. Thien. 1992. Self-incompatibility in Ipomoea pes-caprae (Convolvulaceae). Am. Midl. Nat. 128:22-29.

62

Dormann, C.F., J. Frund, N. Bluthgen, and B. Gruber. 2009. Indices, graphs and null models: analysing bipartite ecological networks. Open J. Ecol. 2:7-24.

Dunne, J. A., R. J. Williams, and N. D. Martinez. 2002 Food-web structure and network theory: the role of connectance and size. PNAS 99:12917– 12922.

Dupont, Y.L., D.M. Hansen, and J.M. Olesen. 2003. Structure of a plant-flower-visitor network in the high altitude sub-alpine desert of Tenerife, Canary Islands. Ecography 26:301-310.

Elmore, M. 2008. Pollination biology of Hawaiian Scaevola (Goodeniaceae). MSc in Botany Dissertation, University of Hawai‘i.

Englund, R.A. 2002. The loss of native biodiversity and continuing nonindigenous species introduction in freshwater, estuarine, and wetland communities of Pearl Harbor, Oah‘u, Hawaiian Islands. Estuaries 25:418-430.

Finkelstein, R.R., S.S.L. Gampala, and C.D. Rock. 2002. Abscisic acid signaling in seeds and seedlings. Plant Cell 14:S15-S45.

Forup, M.L., K.S.E. Henson, P.G. Craze, and J. Memmott. 2008. The restoration of ecological interactions: plant-pollinator networks on ancient and restored heathlands. J. Appl. Ecol. 45:745- 752.

Gagne, W.C. and L.W. Cuddihy. 1999. Vegetation. In: Wagner, W.L., D.R. Herbst, and S.H. Sohmer. 1999. Manual of Flowering Plants of Hawai‘i Revised edition. University of Hawai‘i Press, Honolulu.

Galetto, L., and G. Bernardello. 2004. Floral nectaries, nectar production dynamics, and chemical composition in six Ipomea species (Convolvulaceae) in relation to pollinators. Ann. Bot. 94:269-280.

Gathmann, A., and T. Tscharntke. 2002. Foraging ranges of solitary bees. J. Anim. Ecol. 71:757- 764.

Giambelluca, T.W., Q. Chen, A.G. Frazier, J.P. Price, Y.-L. Chen, P.-S. Chu, J.K. Eischeid, and D.M. Delparte, 2013: Online Rainfall Atlas of Hawai‘i. Bull. Amer. Meteor. Soc. 94, 313-316.

Giambelluca, T.W., X. Shuai, M.L. Barnes, R.J. Alliss, R.J. Longman, T. Miura, Q. Chen, A.G. Frazier, R.G. Mudd, L. Cuo, and A.D. Businger. 2014. Evapotranspiration of Hawaiʽi. Final report submitted to the U.S. Army Corps of Engineers – Honolulu District, and the Commission on Water Resource Management, State of Hawaiʽi.

Goodwillie, C., S. Kalisz, and C.G. Eckert. 2005. The evolutionary enigma of mixed mating systems in plants: Occurrence, theoretical explanations, and empirical evidence. Annu. Rev. Ecol. Evol. Syst. 36:47-79.

63

Gotelli, N.J. and W. Ulrich. 2011. Statistical challenges in null model analysis. Oikos 121:1-10.

Greenleaf, S.S., N.M. Williams, R. Winfree, and C. Kremen. 2007. Bee foraging ranges and their relationship to body size. Oecologia 153:589-596.

Hawaiian Native Plant Propagation Database. 2001. Jacquemontia ovalifolia. College of Tropical Agriculture and Human Resources University of Hawaiʽi at Mānoa.

Hegland, S.J., J. Dunne, A. Nielsen, and J. Memmott. 2010. How to monitor ecological communities cost-efficiently: The example of plant-pollinator networks. Biol. Conserv. 143:2092-2101.

Henneman, M.L., and J. Memmott. 2001. Infiltration of a Hawaiian community by introduced biological control agents. Science 293:1314.

Hopper, D.R. 2002. The reproductive biology and conservation of endangered Hawaiian legume, Sesbania tomentosa, with emphasis on its pollination system. PhD in Zoology Dissertation, University of Hawai‘i.

Howarth, F.G. 1985. Impacts of alien land arthropods and mollusks on native plants and animals in Hawai‘i. In: Stone, C.P., and J.M. Scott, editors. 1985. Hawai‘i’s terrestrial ecosystems preservation and management. Cooperative National park Resources Studies Unit and University of Hawai‘i, Mānoa, HI.

Ings, T.C., J. Schikora, and L. Chittka. 2005. Bumblebees, humble pollinators or assiduous invaders? A population comparison of foraging performance in Bombus terrestris. Oecologia 144:508-516.

Jordano, P. 1987. Patterns of mutualistic interactions in pollination and seed dispersal: connectance, dependence asymmetries, and coevolution. Am. Nat. 129:657-677.

Jordano, P., J. Bascompte, and J.M. Olesen. 2006. The ecological consequences of complex topology and nested structure in pollination webs. In: Waser, N.M. and J. Ollerton, editors. 2006. Plant-pollinator interactions: from specialization to generalization. Chicago, University of Chicago Press. Chapter 8, 173-199.

Junker, R.R., C.C. Daehler, S. Dotterl, A. Keller, and N. Bluthgen. 2011. Hawaiian ant-flower networks: nectar-thieving ants prefer undefended native over introduced plants with floral defenses. Ecol. Monogr. 81:295-311.

Kaiser-Bunbury, C.N., and C.B. Muller. 2009. Indirect interactions between invasive and native plants via pollinators. Naturwissenschaften 96:339-346.

Kaiser-Bunbury, C.N., J. Memmott, and C.B. Muller. 2010a. Community structure of pollination webs of Mauritian heathland habitats. Perspect. Plant Ecol. Evol. Syst. 11:241-254.

64

Kaiser-Bunbury, C.N., A. Traveset, and D.M. Hansen. 2010b. Conservation and restoration of plant-animal mutualisms on oceanic islands. Perspect. Plant Ecol. Evol. Syst. 12:131-143.

Kaiser-Bunbury, C.N., T. Valentin, J. Mougal, D. Matatiken, and J. Ghazoul. 2011. The tolerance of island plant-pollinator networks to alien plants. J. Ecol. 99:202-213.

Kalisz, S., and D.W. Vogler. 2003. Benefits of autonomous selfing under unpredictable pollinator environments. Ecology 84:2928-2942.

Kenta, T., N. Inari, T. Nagamitsu, K. Goka, and T. Hiura. 2007. Commercialized European bumblebee can cause pollination disturbance: an experiment on seven native plant species in Japan. Biol. Conserv. 134:298-309.

Kiers, E.T., T.M. Palmer, A.R. Ives, J.F. Bruno, and J.L. Bronstein. 2010. Mutualisms in a changing world: an evolutionary perspective. Ecol. Lett. 13:1459-1474.

Kiill, L.H., and N.T. Ranga. 2000. Floral biology and breeding system Jacquemontia multiflora (Choisy) Hallier f. (Convolvulaceae). Rev. Bras. Bot. 23:37-43.

Kiill, L.H., and R. Simao-Bianchini. 2011. Reproductive biology and pollination of Jacquemontia nodiflora (Desr.) G. Don (Convolvulaceae) in the Caatinga region of Petrolina, PE, Brazil. Hoehnea 38:511-520.

Koch, J.B., and H.F. Sahli. 2013. Patterns of flower visitation across elevation and successional gradients in Hawaiʽi. Pac. Sci. 67:253-266.

Kratochwil, A., M. Beil, and A. Schwabe. 2009. Complex structure of pollinator-plant interaction-webs: random, nested, with gradients or modules? Apidologie 40:634-650.

Kraus, F., A. Medeiros, D. Preston, C.S. Jarnevich, and G.H. Rodda. 2012. Diet and conservation implications of an invasive chameleon, Chamaeleo jacksonii (Squamata: Chamaeleonidae) in Hawai‘i. Biol. Invasions 14:579-593.

Krushelnycky, P.D., and R.G. Gillespie. 2008. Compositional and functional stability of arthropod communities in the face of ant invasions. Ecol. Appl. 18:1547-1562.

Krushelnycky, P.D. 2014. Evaluating the interacting influences of pollination, seed predation, invasive species and isolation on reproductive success in a threatened alpine plant. PLoS ONE 9:e88948.

Lach, L. 2008. Floral visitation patterns of two invasive ant species and their effects on other hymenopteran visitors. Ecol. Entomol. 33:155-160.

65

LeVan, K., K.L.J. Hung, K.R. McCann, J.T. Ludka, and D.A. Howay. 2014. Floral visitation by the Argentine ant reduces pollinator visitation and seed set in the coast barrel cactus, Ferocactus viridescens. Oecologia 174:163-171.

Liu, F., C.R. Jensen, and M.N. Andersen. 2004. Drought stress effect on carbohydrate concentration in soybean leaves and pods during early reproductive development: its implication in altering pod set. Field Crop Res. 86:1-13.

Lloyd, D.G. 1992. Self- and cross-fertilization in plants. II. The selection of self-fertilization. Int. J. Plant Sci. 153:370-380.

Lloyd, D.G., and D.J. Schoen. 1992. Self- and cross-fertilization in plants. I. Functional Dimensions. Int. J. Plant Sci. 153:358-369.

Lopezaraiza-Mikel, M.E., R.B. Hayes, M.R. Whalley, and J. Memmott. 2007. The impact of an alien plant on a native plant-pollinator network: an experimental approach. Ecol. Lett. 10:539- 550.

MacArthur, R.H., and E.O. Wilson. 1967. The theory of island biogeography. Princeton University Press, Princeton, Massachusetts.

Magnacca, K.N., and B.N. Danforth. 2006. Evolution and biogeography of native Hawaiian Hylaeus bees (Hymenoptera: Colletidae). Cladistics 22:393-411.

Magnacca, K.N. 2007. Conservation Status of the Endemic Bees of Hawai‘i, Hylaeus (Nesoprosopis) (Hymenoptera: Colletidae). Pac. Sci. 61:173-190.

Magnacca, K.N., and C.B.K. King. 2013. Assessing the presence and distribution of 23 Hawaiian yellow-faced bee species on lands adjacent to military installations on Oah‘u and Hawai‘i Island. Technical Report No. 185. Pacific Cooperative Studies Unit, University of Hawai‘i, Honolulu, Hawai‘i. 39 pp.

Maimoni-Rodella, R.C.S., and Y.A.N.P. Yanagizawa. 2007. Floral biology and breeding system of three Ipomea weeds. Planta Daninha 25:35-42.

Maschinski, J., and S.J. Wright. 2006. Using ecological theory to plan restorations of the endangered Beach jacquemontia (Convolvulaceae) in fragmented habitats. J. Nat. Conserv. 14:180-189.

McMullen, C.K. 1987. Breeding systems of selected Galapagos Islands Angiosperms. Am. J. Bot. 74:1694-1705.

McMullen, C.K. 2009. Pollination biology of a night-flowering Galapagos endemic, Ipomoea habeliana (Convolvulaceae). Bot. J. Linn. Soc. 160:11-20.

66

Medeiros, M.J., J.A. Eiben, W.P. Haines, R.L. Kaholoaa, C.B.A. King, P.D. Krushelnycky, K.N. Magnacca, D. Rubinoff, F. Starr, and K. Starr. 2013. The importance of insect monitoring to conservation actions in Hawai‘i. Proc. Hawaiʽi. Entomol. Soc. 45:149-166.

Memmott, J., N.M. Waser, and M.V. Price. 2004. Tolerance of pollination networks to species extinctions. Proc. R. Soc. Lond. B 271:2605-2611.

Namoff, S., Q. Luke, F. Jimenez, A. Veloz, C.E. Lewis, V. Sosa, M. Maunder, and J. Francisco- Ortega. 2010. Phylogenetic analyses of nucleotide sequences confirm a unique plant intercontinental disjunction between tropical Africa, the Caribbean, and the Hawaiian Islands. J. Plant Res. 123:57-65.

Nielson, A., and J. Bascompte. 2006. Ecological networks, nestedness and sampling effort. J. Ecol. 95:1134-1141.

Olesen, J.M., L.I. Eskildsen, and S. Venkatasamy. 2002. Invasion of pollination networks on oceanic islands: importance of invader complexes and endemic super generalists. Divers. Distrib. 8:181-192.

Ollerton, J., R. Winfree, and S. Tarrant. 2011. How many flowering plants are pollinated by animals? Oikos 120:321-326.

Padron, B., A. Traveset, T. Biedenweg, D. Diaz, M. Nogales, and J.M. Olesen. 2009. Impact of alien plant invaders on pollination networks in two archipelagos. PLoS ONE 4:1-11.

Pattemore, D.E., and D.S. Wilcove. 2012. Invasive rats and recent colonist birds partially compensate for the loss of endemic New Zealand pollinators. Proc. R. Soc. B 279: 1597-1605.

Philipp, M., J. Bocher, H.R. Siegismund, and L.R. Nielsen. Structure of a plant-pollinator network on a pahoehoe lava desert of the Galapagos Islands. Ecography 29:531-540.

Pick, R.A., and C. Schlindwein. 2011. Pollen partitioning of three species of Convolvulaceae among oligolectic bees in the Caatinga of Brazil. Plant Syst. Evol. 293:147-159.

Pimental, D., R. Zuniga, and D. Morrison. 2005. Update on the environmental and economic costs associated with alien-invasive species in the United States. Ecol. Econ. 52:273-288.

Pinto-Torres, E., and S. Koptur. 2009. Hanging by a coastal strand: breeding system of a federally endangered morning-glory of the south-eastern Florida coast, Jacquemontia reclinata. Ann. Bot. 104:1301-1311.

Pleasants, J.M., and J.F. Wendel. 2010. Reproductive and pollination biology of the endemic Hawaiian cotton, Gossypium tomentosum (Malvaceae). Pac. Sci. 64:45-55.

Popic, T.J., G.M. Wardle, and Y.C. Davila. 2013. Flower-visitor networks only partially predict the function of pollen transport by bees. Austral Ecol. 38:76-86.

67

Potts, S.G., J.C. Biesmeijer, C. Kremen, P. Neumann, O. Schweiger, and W.E. Kunin. 2010. Global Pollinator Declines: trends, impacts and drivers. Trends Ecol. Evol. 25:345-53.

Pratt, L. W., J. R. VanDeMark, and M. Euaparadorn. 2011. Status and limiting factors of three rare plant species in the coastal lowlands and mid-elevation woodlands of Hawai`i Volcanoes National Park. Hawai‘i Cooperative Studies Unit Technical Report HCSU-024. University of Hawai‘i at Hilo. 89 pp., incl. 44 figures, 5 tables, & 2 appendices.

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

Raimundez-Urrutia, E., L. Avendano, and D. Velazquez. 2008. Reproductive biology of the morning glory Merremia macrocalyx (Ruiz & Pavon) O’Donnell (Convolvulaceae). J. Torrey Bot. Soc. 135:299-308.

Richmond, T.A., and D. Mueller-Dombois. 1972. Coastline ecosystems on Oah‘u, Hawai‘i. Vegetatio 25:367-400.

Robertson, K.R. 1974. Jacquemontia ovalifolia (Convolvulaceae) in Africa, North America, and the Hawaiian Islands. Ann. Missouri Bot. 61:502-513.

Sakai, A.K., W.L. Wagner, D.M. Ferguson, and D.R. Herbst. 1995. Origins of dioecy in the Hawaiian flora. Ecology 76:2517-2529.

Schlessman, M.A., L.B. Vary, J. Munzinger, and P.P. Lowry II. 2014. Correlates, and origins of dioecy in the island flora of New Caledonia. Int. J. Plant Sci. 175:271-286.

Schoen, D.J., and D.G. Lloyd. 1992. Self- and cross-fertilization in Plants. III. Methods for studying modes and functional aspects of self-fertilization. Int. J. Plant Sci. 153:381-393.

Silva, F.O., S.D. Kevan, N. Roque, B.F. Viana, and P.G. Kevan. 2010. Records on floral biology and visitors of Jacquemontia montana (Moric.) Meisn. (Convolvulaceae) in Mucuge, Bahia. Braz. J. Biol. 70:671-676.

Su, A., X. Ma, H. Guo, N.L. Sukiran, B. Guo, S.M. Assmann, and H. Ma. 2013. Flower development under drought stress: Morphological and transcriptomic analyses reveal acute responses and long-term acclimation in Arabidopsis. Plant Cell 25:3785-3807.

Suarez, L.H., W.L. Gonzalez, and E. Gianoli. 2004. Reproductive biology of Convolvulus chilensis (Convolvulaceae) in a population of Auco (north-central Chile). RCHN. 77:581-591.

Traveset, A., and D.M. Richardson. 2006. Biological invasions as disruptors of plant reproductive mutualisms. Trends Ecol. Evol. 21:208-216.

68

Traveset, A., R. Heleno, S. Chamorro, P. Vargas, C.K. McMullen, R. Castro-Urgal, M. Nogales, H.W. Herrera, and J.M. Olesen. 2013. Invaders of pollination networks in the Galapagos Islands: emergence of novel communities. Proc. R. Soc. B 280:20123040.

Trojelsgaard, K., and J.M. Olesen. 2013. Macroecology of pollination networks. Global Ecol. Biogeogr. 22:149-162.

Tylianakis, J.M., T. Tscharntke, and O.T. Lewis. 2007. Habitat modification alters the structure of tropical host-parasitoid food webs. Nature 445:202-205.

Tylianakis, J.M., E. Laliberte, A. Nielsen, and J. Bascompte. 2010. Conservation of species interaction networks. Biol. Conserv. 143:2270-2279.

Ulrich, W., M. Almeida-Neto, and N.J. Gotelli. 2009. A consumer’s guide to nestedness analysis. Oikos 118: 3-17.

Ushimaru, A., and K. Kikuzawa. 1999. Variation of breeding system, floral rewards, and reproductive success in clonal Calystegia. Am. J. Bot. 86:436-446.

Vila, M., I. Bartomeus, A.C. Dietzsch, T. Petanidou, I. Steffan-Dewenter, J.C. Stout, and T. Tscheulin. 2009. Invasive plant integration into native plant-pollinator networks across Europe. Proc. R. Soc. B 276:3887-3893.

Wagner, W.L., D.R. Herbst, and S.H. Sohmer. 1999. Manual of Flowering Plants of Hawaiʽi Revised edition. University of Hawaiʽi Press, Honolulu, HI.

Warshauer, F. R., J. D. Jacobi, and J. Price 2009. Native coastal flora and plant communities in Hawai‘i: Their composition, distribution, and status. Hawai‘i Cooperative Studies Unit Technical Report HCSU-014. University of Hawai‘i at Hilo.

Weller, S.G., A.K. Sakai, A.E. Rankin, A. Golonka, B. Kutcher, and K.E. Ashby. 1998. Dioecy and the evolution of pollination systems in Schidea and Alsinidendron (Caryophyllaceae: Alsinoideae) in the Hawaiian Islands. Am. J. Bot. 85:1377-1388.

Whittaker, R.J. and J.M. Fernandez-Palacios. 2007. Island Biogeography: Ecology, evolution, and conservation. Oxford University Press Inc., New York.

Willmer, P. 2011. Pollination and floral ecology. Princeton University Press, Princeton.

Wilson, E.E., C.S. Sidhu, K.E. LeVan, and D.A. Holway. 2010. Pollen foraging behaviour of solitary Hawaiian bees revealed through molecular pollen analysis. Mol. Ecol. 19:4823-4829.

Wright, M.A.R., M.D. Ianni, and M. Costea. Diversity and evolution of pollen-ovule production in Cuscuta (dodders, Convolvulaceae) in relation to floral morphology. Plant Syst. Evol. 298:369-389.

69

Yorkston, M., and C.C. Daehler. 2006. Interfertility between Hawaiian ecotypes of Sida fallax (Malvaceae) and evidence of a hybrid disadvantage. Int. J. Plant Sci. 167:221-230.

Young, L.C., E.A. VanderWerf, M.T. Lohr, C.J. Miller, A.J. Titmus, D. Peters, and L. Wilson. 2013. Multi-species predator eradication within a predator-proof fence at Kaʽena Point, Hawai‘i. Biological Invasions 15:2627-2638.

Yorkston, M., and C.C. Daehler. 2006. Interfertility between Hawaiian ecotypes of Sida fallax (Malvaceae) and evidence of a hybrid disadvantage. Int. J. Plant Sci. 167:221-230.

Ziegler, A.D. 2002. Hawaiian Natural History. University of Hawai‘i Press, Honolulu.

Zimmerman, E.C. 1958. Macrolepidoptera. Insects of Hawai‘i, vol. 7. University of Hawai‘i Press, Honolulu.

70