THE EVOLUTIONARY HISTORY of HAWAIIAN LEAF-MINING MOTHS in the GENUS Philodoria (GRACILLARIIDAE: LEPIDOPTERA)

Home , Argyroxiphium, Argyroxiphium grayanum, Conopomorpha, Diospyros sandwicensis, Dubautia, Glyphipterigidae

THE EVOLUTIONARY HISTORY OF HAWAIIAN LEAF-MINING MOTHS IN THE GENUS Philodoria (GRACILLARIIDAE: LEPIDOPTERA)

CHRISTOPHER AGUSTIN JOHNS

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

UNIVERSITY OF FLORIDA

2017

To my friends, family, and the Islands

ACKNOWLEDGMENTS

I thank Charmian Dang, Betsy Gangé, and Cynthia King (Hawai‘i Department of

Land and Natural Resources, Division of Forestry and Wildlife) for permitting; Keahi M.

Bustamente (Leeward Haleakalā Watershed Restoration Project) and Natalia Tangalin

(National Tropical Botanical Garden) for endangered plant identification, collection assistance, reports on field observations, and project planning; Margaret J. Sporck-

Koehler (Hawai‘i Department of Land and Natural Resources) for assistance with fieldwork, land access, and gathering information on host plant conservation statuses;

Pat Bily and Russel Kalstrom (Nature Conservancy) for land access; Pomaika‘i

Kanaiaupio-Crozier, Lono Dunn, Daniel Tanaka, and Joe Ward (Pu‘u Kukui Watershed

Preserve) for land access and field support; West Maui Mountain Watershed

Partnership for logistical support; Avery Chumbley for West Maui access to land; Butch

Haase (Moloka‘i Coastal Land Trust), Ane Bakutis (Hawai‘i Plant Extinction Prevention

Program) for support of fieldwork on Moloka‘i; William P. Haines (Univ. of Hawai‘i,

Mānoa), Karl Magnacca (Oahu Army Natural Resources Program), Hank Oppenheimer

(Hawai‘i Plant Extinction Prevention Program), Susan Ching Harbin (Hawai‘i Plant

Extinction Prevention Program), Joel Q. C. Lau, and Steve Montgomery for helpful suggestions on insect and host plant locations; Jesse Eiben (University of Hawai‘i Hilo),

Mark Wasser (Hawai‘i Volcanoes National Park), and Melissa Dean (Hawai‘i

Experimental Tropical Forest) for help coordinating fieldwork on Hawai‘i Island; Diana

Crow (Ulupalakua Ranch) for assistance with access on East Maui; Marie VanZandt

(Auwahi Wind Energy) and Fritz Klasner (Office of Mauna Kea Management) for permitting and access; Daniel Rubinoff (Univ. of Hawai‘i, Mānoa) for the initial development of the project; Sheperd Meyers and Barbara Kennedy for allowing

examination of insect and herbarium specimens at the Bernice Pauahi Bishop Museum;

Donald R. Davis (Smithsonian National Museum) and Shigeki Kobayashi (Osaka

Prefecture University) for assisting with specimen identification; Jesse W. Breinholt,

Emmanuel F.A. Toussaint, David Plotkin, Francesca Ponce, Kelly Dexter, and Lei Xiao

(Florida Museum of Natural History) for assistance with lab work and data analysis; and

Jonathan Bremer (Florida Museum of Natural History) for assistance in herbarium analyses.

Special thanks to Thomas C. Emmel (McGuire Center for Lepidoptera and

Biodiversity) for guidance and early support of the project; University of Florida

Department of Entomology for initial support of the research; faculty and staff of the

McGuire Center for Lepidoptera and Biodiversity and the members of the Kawahara Lab for support and collaboration; Nico Cellinese, Gustav Paulay, and Jiri Hulcr (University of Florida) for their support of my academic and personal development; and the attendants of the 2015 (Japan) and 2016 (Hawai‘i) International Symposium on

Gracillariidae for their comments and direction on research questions.

This research was supported by the National Science Foundation (Graduate

Research Fellowship to C. A. J.; DEB #1354585 to A. Y. K); the National Geographic

Society (#C283-14 to C. A. J., and #9686-15 to A. Y. K.); the Entomological Society of

America (2014 SysEB Travel Award to C. A. J); the University of Florida’s Tropical

Conservation and Development Program (2014 Field Research Grant to C. A. J); the

Florida Museum of Natural History (Lockhart Fellowship and Travel Award to C. A. J.); the International Biodiversity Foundation (Documentary Production Grant to C. A. J.); and the Society for Systematic Biologists’ (2012 SSB MiniARTS Grant to A. Y. K).

This body of work would not have been possible without the support of Thomas

Johns, Elizabeth Agustin, DeeAnn Johns, Ryan Johns, Sophia Ramos, Geena Hill, Rick

Merker, Emily Johns, Juli Burden, Niko Marshall, Parker Pflaum, Kai Moreb, Keahi

Bustamente, Nora Beale, and Calvin Beale. Lastly, I thank Akito Y. Kawahara, for his tremendous friendship, guidance, and support.

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 9

LIST OF FIGURES ...... 10

LIST OF OBJECTS ...... 11

ABSTRACT ...... 12

CHAPTER

1 MOLECULAR PHYLOGENY, REVISED HIGHER CLASSIFICATION, AND IMPLICATIONS FOR THE CONSERVATION OF ENDANGERED HAWAIIAN LEAF-MINING MOTHS (LEPIDOPTERA: GRACILLARIIDAE: Philodoria) ...... 13

Introduction ...... 13 Materials and Methods ...... 17 Taxon Sampling, Amplification, and Sequencing ...... 17 Phylogenetic Analyses ...... 20 Hypothesis Testing ...... 21 Results ...... 23 Discussion ...... 23 Subgeneric Classification ...... 23 Philodoria: Implications for Ecology and Conservation ...... 25

2 EVIDENCE OF AN UNDESCRIBED, EXTINCT Philodoria SPECIES (LEPIDOPTERA: GRACILLARIIDAE) FROM HAWAIIAN Hesperomannia HERBARIUM SPECIMENS ...... 32

3 CONSERVATION MEDIA AS A TOOL TO RAISE AWARENESS FOR Philodoria MOTHS AND THEIR CONSERVATION ...... 36

Background ...... 36 Engagement ...... 38 Future Considerations ...... 38

4 EVOLUTIONARY HISTORY OF Philodoria ...... 40

Introduction ...... 40 Hawaiian Geology ...... 40 Origin of the Hawaiian Biota ...... 42 Calibration Strategy ...... 42

Philodoria ...... 44 Materials and Methods ...... 45 Taxon Sampling ...... 45 Sample Preparation, Sequencing, Data Processing, and Dataset Construction ...... 51 Phylogenomic Analyses ...... 53 Divergence Time Estimation ...... 54 Historical Biogeography and Ancestral Host Plant Reconstruction ...... 57 Phylogenomics ...... 59 Divergence Time Estimation ...... 64 Historical Biogeography and Ancestral Host Plant Association ...... 65 Discussion ...... 70 Phylogenomics ...... 70 Dataset Subsampling ...... 71 Calibration Strategy ...... 72 Evolution of Philodoria and Their Host Associations ...... 74 Conservation of Philodoria Moths and Their Host Plants ...... 80

APPENDIX

A ACCESSION AND PARTITION TABLES ...... 81

B PARTITION TABLES ...... 82

LIST OF REFERENCES ...... 86

BIOGRAPHICAL SKETCH ...... 94

LIST OF TABLES

Table page

1-1 Summary of current Philodoria classification, diversity, and host plants of taxa included in this study...... 15

1-2 Primer, primer nucleotide sequence, gene, size, and author of sequences used in phylogenetic analyses of Philodoria...... 19

1-3 Host plant genera mined by Philodoria moths and the conservation status of their members...... 29

4-1 Philodoria species sequenced in this study, their host plants, and localities from which they were sampled...... 49

4-2 Divergence time analysis results, with ages in millions of years with 95% highest posterior density range in parentheses...... 63

4-3 BioGeoBEARS results. Parameters = d (dispersal), e (extinction), j (jump dispersal), and x (dispersal modified by distance). LRT values compare pairs of models with nested parameters...... 67

A-1 GenBank accession numbers for ingroup and outgroup taxa...... 81

A-2 Data set partitions and the corresponding best-fitting model of sequence evolution...... 81

B-1 Dataset 1 partition models. Asterisk next to partition number indicates most likely model...... 82

B-2 Dataset 2 partition models. Asterisk next to partition number indicates most likely model...... 83

LIST OF FIGURES

Figure page

1-1 Map of the Hawaiian Islands and the collection localities for the taxa sampled in this study. Additional information is available in Table 1-1...... 18

1-2 Majority rule consensus tree from BI analyses of Philodoria. Host plant, collection locality, and subgeneric classification are also displayed...... 22

2-1 Philodoria pupal tents on the adaxial surface of Hesperomannia arborescens leaves from Lanai (BISH1022034)...... 34

4-1 Map of collecting sites visited in the course of this study. Map produced in Google Earth Pro...... 46

4-2 ALISTAT heatmap of pairwise nucleotide completeness for the final nucleotide alignment of Dataset 1...... 60

4-3 Maximum likelihood tree of Philodoria estimated from Dataset 2 (full-length loci), with the support values from both datasets including Dataset 1 (probe region only) and from both bootstrapping methods included for comparison...... 61

4-4 Violin plots of Philodoria crown age from most likely divergence time analyses...... 62

4-5 MCC chronogram from BEAST, with nodes ages in millions of years and blue bars at nodes representing the 95% highest posterior density range...... 66

4-6 Ancestral state reconstruction of Philodoria host plant associations using the MK-1 model...... 68

4-7 Historical biogeography of Philodoria inferred by BioGeoBEARS. Areas are as follows: N = Nihoa, K = Kauai, O = Oahu, M = Maui Nui, H = Hawaii Island...... 69

B-1 ASTRAL tree. Red numbers show ASTRAL bootstrap support. Blue boxes indicate full support...... 84

B-2 ML tree topology from 50-locus subsampled Robinson-Foulds dataset...... 85

LIST OF OBJECTS

Object page

3-1 LEAF MINERS short film (.mp4 file 1253 MB) ...... 38

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

THE EVOLUTIONARY HISTORY OF HAWAIIAN LEAF-MINING MOTHS IN THE GENUS Philodoria (GRACILLARIIDAE: LEPIDOPTERA)

Christopher Agustin Johns

December 2017

Chair: Akito Y. Kawahara Major: Zoology

The endangered leaf mining moth genus Philodoria Walsingham 1907

(Lepidoptera: Gracillariidae) comprises 44 species, all of which are endemic to the

Hawaiian Islands. Most Philodoria species are monophagous, but the genus as a whole is known to feed within the leaf tissue of six families of endemic Hawaiian host plants.

Several members of the group are closely associated with iconic endemic plant lineages, but their complete host plant range and classification remain unclear. Most of the known Philodoria host plant genera include threatened or endangered plant species.

Despite their unique life history and threatened existence, Philodoria has received little scientific or conservation attention since its initial description over a century ago. The present study examines the evolutionary history of Philodoria, including their colonization of the Hawaiian Islands, the evolution of their host range, and their dependence on threatened and endangered plants. A significant component of this body of work also focuses on raising awareness for Philodoria and Hawaiian insect conservation through documentary film.

CHAPTER 1 MOLECULAR PHYLOGENY, REVISED HIGHER CLASSIFICATION, AND IMPLICATIONS FOR THE CONSERVATION OF ENDANGERED HAWAIIAN LEAF- MINING MOTHS (LEPIDOPTERA: GRACILLARIIDAE: Philodoria)

Introduction

Philodoria Walsingham 1907 (Lepidoptera: Gracillariidae) is a genus of leaf- mining micromoths that currently includes 30 described species, all of which are endemic to the Hawaiian Islands (Zimmerman 1978). The genus is extraordinary in that its larvae mine leaves of ten plant families from seven orders (Swezey 1954;

Zimmerman 1978). Host plant groups include iconic and endangered Hawaiian plant taxa such as the silversword alliance (Argyroxiphium DC. and Dubautia Gaudich.) and the Hawaiian lobelioids (Clermontia Gaudich.). Approximately 80% of Philodoria species feed on a single plant host species, and more than three-quarters of these species are restricted to a single Hawaiian island (Zimmerman 1978). The genus Philodoria should be considered a conservation priority due to the stringent host specificity and limited geographic range of the majority of its species.

The taxonomic history of Philodoria has been unstable and the group’s evolutionary relationships remain unknown despite its distribution across Hawai‘i and specialization on distantly related plants. Philodoria was originally assigned to Tineidae

Latreille (Walsingham 1907), followed by placement in Glyphipterigidae Stainton

(Meyrick 1912). Species within Philodoria have also been assigned to various other genera, including Gracillaria Haworth, Elachista Treitschke (Walsingham 1907), and

Parectopa Clemens (Meyrick 1928). The most recent systematic treatment grouped all

Hawaiian species previously assigned to Elachista, Gracillaria, and Parectopa into

Philodoria (Zimmerman 1978). Zimmerman divided the genus into two subgenera,

Philodoria (Eophilodoria) and Philodoria (Philodoria), based on the size of the maxillary palpus. Under this classification, Zimmerman assigned 16 Philodoria species with the maxillary palps “fully developed” to the subgenus Eophilodoria (type species: P. marginestrigata Walsingham). Fourteen Philodoria species with this structure “greatly reduced, vestigial, or obsolescent” were assigned to the subgenus Philodoria (type species: P. succedanea Walsingham). In addition, Zimmerman’s treatment defines

Philodoria species based on scale patterns, host plant associations, and distribution.

However, no phylogenetic data/analyses have evaluated the usefulness of these characters for defining the subgenera or species. This study represents the first attempt to evaluate the usefulness of the maxillary palp character (i.e., the monophyly of the subgenera) for the subgeneric classification of Philodoria.

We constructed the first phylogeny of Philodoria that sampled molecular sequence data from one mitochondrial and two nuclear genes from 11 Philodoria species (Table 1-1) in order to test the subgeneric classification of Zimmerman (1978).

Our results do not support Zimmerman’s groupings and we discuss patterns of host plant associations among our sampled Philodoria species.

Table 1-1. Summary of current Philodoria classification, diversity, and host plants of taxa included in this study.

Last Study Host plant Host plant documented Moth Collection locality Moth taxon Author Subgenus specimen species1 family previous to Dist.1,3 (DDD.DDDDD°) ID this study1

Walsingham Metrosideros H, K, L4, 21.504618° N, - P. splendida Philodoria Myrtaceae 1943 CJ-049 1907 polymorpha Mo, O5 158.146989° W

(Meyrick Melanthera 21.572609° N, - P. sciallactis Eophilodoria Asteraceae 1927 CJ-061 O4,5 1928) integrifolia 158.275253° W

Myrsine Walsingham lessertiana; H, M5✝, 20.930785° N, P. auromagnifica Philodoria Primulaceae 1928 CJ-064 1907 Myrsine Mo, O4 156.610357° W sandwicensis2

(Swezey Hibiscus H, K, 20.881126° N, - P. hauicola Eophilodoria Malvaceae 1910 CJ-065 1910) tiliaceus M5, O4 156.546767° W

Argyroxiphium 20.910519° N, - P. wilkesiella Swezey 1940 Philodoria Asteraceae 1940 CJ-068 M4, 5 grayanum 156.592075° W

Walsingham Myrsine H, M5✝, 20.930585° N, - P. auromagnifica Philodoria Primulaceae 1928 CJ-072 1907 lessertiana Mo, O4 156.610279° W

Dubautia (Swezey 20.911807° N, - P. dubauticola Eophilodoria plantaginea; Asteraceae 1940 CJ-077 M4,5 1940) 156.592131° W Dubautia laxa2

(Swezey 21.412447° N, - P. pipturicola Philodoria Pipturus sp. Urticaceae 1928 CJ-101 M, O4,5 1915) 158.100055° W

Table 1-1. Continued

Last Study Collection Host plant Host plant documented Moth Moth taxon Author Subgenus specimen locality species1 family previous to Dist.1,3 ID (DDD.DDDDD°) this study1 H, K, L4, Walsingham Metrosideros 20.976205° N, - P. splendida Philodoria Myrtaceae 1919 CJ-105 M5✝, 1907 polymorpha 156.61929° W Mo, O Walsingham Metrosideros 20.818547° N, - P. basalis Philodoria Myrtaceae 1919 CJ-112 H, M4,5 1907 polymorpha 156.268527° W H, K, Walsingham Abutilon; Asteraceae; 21.572609° N, - P. marginestrigata* Eophilodoria 1990 CJ-135 Mo, N, 1907 Dubautia; Sida Malvaceae 158.275253° W O4,5 Dubautia laxa; (Swezey 21.533236° N, - P. naenaeiella Eophilodoria Dubautia Asteraceae 1943 CJ-142 M, O4,5 1940) 157.927446° W plantaginea2 Myrsine Walsingham sandwicensis; 20.785926° N, - P. succedanea* Philodoria Primulaceae 1896 CJ-144 H, M4,5 1907 Myrsine 156.230492° W lessertiana2 Notes: 1As indicated by pinned (BPBM and USNM) and published (Zimmerman 1978) data. 2New host plant records observed in this study. 3Islands: H, Hawai‘i; K, Kaua‘i; L, Lāna‘i; M, Maui; Mo, Moloka‘i; N, Ni‘ihau; O, O‘ahu. 4Island of type locality. 5Collection locality of associated specimen. *Type species of corresponding subgenus. ✝New island record.

Materials and Methods

Taxon Sampling, Amplification, and Sequencing

Thirteen samples representing 11 species of Philodoria were collected during

April 2013 at 13 sites on the islands of O‘ahu and Maui (Figure 1-1). The type species of each of the subgenera defined by Zimmerman (1978), Philodoria (Eophilodoria) marginestrigata and Philodoria (Philodoria) succedanea, were captured in these collections (Table 1-1). Philodoria collection localities were selected based on historical records of Swezey (1954) and Zimmerman (1978). New localities were also surveyed based on the presence of known Philodoria host plant species. We visually identified host plants and collected leaves with signs of leaf miner larval activity. Both inactive and active leaf mines were photographed and georeferenced. Leaves with active mines and advanced larval instars were collected and kept in cool, dry conditions in plastic containers for rearing. Successfully reared moths were stored in 100% ethanol for molecular analyses. Larvae that did not successfully pupate and emerge as adults were stored in ethanol for future morphological and molecular analyses. Moths and the leaves from which they were reared were kept as voucher material and are deposited at the McGuire Center for Lepidoptera and Biodiversity (MGCL), Florida Museum of

Natural History, Gainesville, Florida, USA. Parasitoids reared from these collections are also stored at MGCL.

Figure 1-1. Map of the Hawaiian Islands and the collection localities for the taxa sampled in this study. Additional information is available in Table 1-1.

Multiple representatives of two species (Philodoria auromagnifica – samples CJ-

064, and CJ-072; Philodoria splendida – samples CJ-049, and CJ-105) were included in the study to determine genetic variation between samples collected from different volcanoes or host plants. All adult moths sequenced in this study were reared from active leaf mines as detailed above, with the exception of CJ-049, which was field collected as an adult. Philodoria species were identified by comparing adult morphology with specimens determined by Otto H. Swezey or Elwood C. Zimmerman that were stored in the Bishop Museum, Honolulu (BPBM) or the Smithsonian National Museum,

Washington, D.C. (NMNH). We also aided our identifications by comparing our locality data and larval host plant data with historical records.

Table 1-2. Primer, primer nucleotide sequence, gene, size, and author of sequences used in phylogenetic analyses of Philodoria. Primer Primer nucleotide sequence Gene Size Citation (Hebert et LepF1 ATTCAACCAATCATAAAGATAT CO1 687 al. 2004) LepR1 TAAACTTCTGGATGTCCAAAAA CO1 - (Monteiro ef44 GCYGARCGYGARCGTGGTATYAC EF-1α ~1,100 and Pierce 2001) efrcM4 ACAGCVACKGTYTGYCTCATRTC EF-1α -

TAATACGACTCACTATAGGG[TGGAARG (Kawahara CADm5F CAD 1,056 ARGTNGARTAYGARGT] et al. 2013)

ATTAACCCTCACTAAAG[ACNGCRCACCA CADm1mR CAD - RTCRAAYTCNACNGA] Note: The bracketed portion of CAD denotes the complementary region of the primer to the gene’s sequence.

Molecular data were obtained by extracting the DNA from the entire adult moth.

Extraction methods followed manufacturer’s protocols for the Qiagen DNEasy kit

(Qiagen, Inc., Valencia, CA, U.S.A.). Specimens were sequenced for three genes:

mitochondrial Cytochrome c Oxidase subunit 1 (CO1; 636 bp), nuclear

Carbamoylphosphate Synthase domain of CAD (870 bp), and nuclear Elongation factor

1-alpha (EF-1α; 963 bp); the primer sequences for amplification of each fragment are listed in Table 1-2. We included the same loci for three gracillariids, Epicephala relictella, Parectopa robiniella, and Conopomorpha sp. from the study of Kawahara et al. (2011). These taxa were included as outgroups because they are known to be close relatives of Philodoria (Kawahara et al. 2017). Sequences were edited using Geneious

Pro v5.5.8 (Biomatters 2013) and sequence alignments were produced using the

MUSCLE alignment algorithm with default parameters (Edgar 2004). Each gene alignment was manually concatenated together into a single alignment that totaled

2,041 base pairs. Supplementary Table A-1 lists GenBank accession numbers; the single gene trees, concatenated data set, and photos of sequenced tissue are available from the Dryad data depository (http://datadryad.org).

Phylogenetic Analyses

Analyses using parsimony (P), maximum likelihood (ML), and Bayesian inference

(BI) were first conducted on individual loci to assess congruence among data sets.

Parsimony analyses were executed in PAUP* 4.0 (Swafford 2003) using heuristic searches performed with 1,000 random addition replicates and tree bisection- reconnection (TBR) branch swapping. For ML and BI, we first partitioned the concatenated data set by gene region and codon position, and determined the best- fitting models of sequence evolution for each partition in PartitionFinder 1.0.1 (Lanfear et al. 2012) using the Akaike Information Criterion. The models for each partition were used in the following analyses and are listed in Supplementary Table A-2. ML analyses were implemented in RAxML 8.1.12 (Stamatakis 2014), with 1,000 bootstrap replicates.

Bayesian analyses were conducted in MrBayes 3.2.1 (Ronquist et al. 2012), sampling

MCMC chains every 1,000th tree for 20 million generations. Stationarity was evaluated in Tracer 1.6.0 (Drummond et al. 2012) and 2,000,000 generations (10%) were subsequently discarded as burn-in. No strongly supported topological incongruence was observed between individual gene trees using these methods and identical tree topologies were obtained for each locus. The above parameters were then used to analyze the concatenated data set. Phylogenetic trees were visualized in FigTree 1.4.2

(Rambaut and Drummond 2009).

Hypothesis Testing

In order to compare the confidence between our results and Zimmerman’s (1978) morphology-based hypothesis, we conducted an analysis in which the two Philodoria subgenera were each constrained to be monophyletic. In RAxML, an ML tree was estimated with this constraint enforced and the likelihood score of this tree was compared to the ML tree obtained from the unconstrained analysis. Statistical comparisons between these trees were made with the Shimodaira-Hasegawa (SH) test implemented in RAxML, and the Approximately UnBiased (AU) test of Shimodaira

(2002) implemented in CONSEL 0.20 (Shimodaira and Hasegawa 2001). For the AU test, we estimated site likelihoods for both constrained and unconstrained analyses with

PAUP* (Swafford 2003) before combining them together into a single file for CONSEL.

Figure 1-2. Majority rule consensus tree from BI analyses of Philodoria. Host plant, collection locality, and subgeneric classification are also displayed. Support values indicate bootstrap values for P, followed by bootstraps for ML, and the posterior probabilities from BI, pertaining to the adjacent node. Our results support the division of Philodoria into two clades (referred to as “Clade A” and “Clade B”), but differ from the morphology-based classification proposed by Zimmerman (1978).

Results

Sequencing, Phylogenetic Analysis, and Hypothesis Testing: Nuclear and mitochondrial sequences were obtained from 13 gracillariid specimens. Successful amplifications of all genes were obtained for all but one sample (P. naenaeiella, CJ-142,

CAD), and sequence data for CO1 were missing for one outgroup (Conopomorpha sp.).

The final data matrix had only 4.8% missing data, and individual gene trees had nearly identical ingroup relationships (topological discrepancies were caused by missing data for the two taxa listed above). The concatenated data set resulted in trees with entirely congruent ingroup topologies in all subsequent phylogenetic analyses, regardless of optimality criterion.

Parsimony heuristic searches resulted in one most parsimonious tree (length =

1570, CI = 0.7312, RI = 0.6837). ML and BI analyses also resulted in trees with the same topology as the parsimony tree and branch support was strong (> 70% bootstrap for ML, >0.90 PP for BI) for nearly all nodes. All concatenated trees supported the division of the genus into two main clades (Clade A and B), the composition of which was Clade A (Eophilodoria + Philodoria) and Clade B (remaining Philodoria; Figure 1-2).

Monophyly of subgenera, as previously defined (Zimmerman 1978), was statistically rejected (P < 0.0001) for both SH and AU tests.

Discussion

Subgeneric Classification

Our molecular phylogeny of Philodoria does not support the morphology-based classification of Zimmerman (1978), who split the genus into two subgenera, Philodoria

(Eophilodoria) and Philodoria (Philodoria), based on the development of the maxillary palpus. Zimmerman (1978) classified Philodoria species with a fully developed maxillary

palpus as subgenus Eophilodoria, and species with this structure “greatly reduced, vestigial, or obsolescent” as belonging to subgenus Philodoria. Our results confirm that

Philodoria are classified into two groups (Figure 1-2). Shimodaira-Hasegawa and AU tests statistically rejected the monophlyly of the Zimmerman’s subgenera, as Philodoria wilkesiella and P. pipturicola, species originally described within subgenus Philodoria, were consistently nested within the clade containing species belonging to subgenus

Eophilodoria (Figure 1-2).

Host plant data corroborate the grouping of P. wilkesiella and P. pipturicola with the related taxa found in Clade A (Figure 1-2), suggesting that host ranges for

Philodoria species may be conserved at the level of plant family. Philodoria wilkesiella feeds on the endemic aster, Argyroxiphium grayanum (Hillebr.) O.Deg., and all sampled species that feed on asters were placed in Clade A. Philodoria pipturicola feeds on plant species in the Hawaiian nettle genus Pipturus Wedd. (Urticaceae), which is host to seven Philodoria species across the Hawaiian Islands. All Philodoria species that mine leaves of Pipturus are currently placed within the subgenus Philodoria (Zimmerman

1978). We postulate that the six other Pipturus miners, which were not sampled in this study, are probably incorrectly classified, because they share similarities in morphology and host plant preference with the Pipturus miner included in our analyses. Following this pattern, it is likely that the Philodoria species that mine other Hawaiian plant genera within Urticaceae (Neraudia Gaudich., Touchardia Gaudich., and Urera Gaudich.) are also incorrectly classified. At present, Zimmerman’s (1978) classification places

Philodoria species that feed on these three plant genera in Clade A. Future

phylogenetic studies should include additional Philodoria species to comprehensively elucidate evolutionary patterns of host plant shifts.

The phylogeny reported here suggests that the morphological characters used to classify the Philodoria subgenera may be homoplasious. These data suggest the reduction of the maxillary palps is not a reliable character for the subgenus Philodoria

(Philodoria) or that the interpretation of these characters was otherwise flawed (i.e.,

Zimmerman (1978) based his classification on a compound character or an inadequately defined continuous character). The two main clades A and B (Figure 1-2) recovered in these analyses are well supported and could be treated as revised subgenera. However, there are no reliable morphological characters or hypothesized synapomorphies to separate Zimmerman’s (1978) subgenera for identification purposes, and no obvious ecological differences that define the two main clades in our study. Therefore, the subgeneric rank is here removed. Philodoria Walsingham, 1907 is the oldest name, and the genus-group name Eophilodoria is here placed in synonymy with it (Eophilodoria Zimmerman, 1978, syn.nov.). Results from the present study confirm our poor understanding of Philodoria and demonstrate the need for a closer look at their phylogenetic relationships, current distributions, and conservation status of these species.

Philodoria: Implications for Ecology and Conservation

We present new host plant and distribution data that have implications for the ecology and conservation of Philodoria. Two species that mine Myrsine L. (Ericales:

Primulaceae) were collected in this study. Philodoria succedanea (CJ-144) was reared from Myrsine lessertiana (a previously unrecorded host plant species for this moth species) near to the moth’s type locality on East Maui (Table 1-1). On West Maui, P.

auromagnifica (CJ-064 and CJ-072) was reared from leaves of M. sandwicensis and M. lessertiana. While these specimens very closely resemble P. auromagnifica – a Myrsine miner known from O‘ahu, Moloka‘i, and Hawai‘i Island (Table 1-1) – they differ subtly in wing pattern and may represent an undescribed species. Zimmerman (1978) hypothesized that there might be numerous undescribed species of Philodoria on

Myrsine. Additionally, some Philodoria specimens dissected by Zimmerman and housed at NMNH include label details that indicate he believed they represent new species collected from Myrsine. There are no Myrsine-feeding Philodoria recorded from Kaua‘i, suggesting that there is a gap in host plant sampling on this island, especially considering that Kaua‘i is home to eight described Philodoria species and at least ten

Myrsine species (Wagner et al. 1999), three of which are endangered (U.S. Fish and

Wildlife Service 2015). Future efforts to collect Philodoria on Kaua‘i should focus on

Myrsine species.

Eight Philodoria species feed on Asteraceae and nearly all of these aster feeders are recorded to mine only one host plant genus (Zimmerman 1978). The exception to this pattern is P. marginestrigata (included in this study), which is recorded to mine plants in Asteraceae and Malvaceae. The association of this moth species with

Asteraceae, however, remains dubious, and it is likely that early observations of an aster host plant for P. marginestrigata were incorrect (Zimmerman 1978).

The remaining aster-feeding Philodoria species are all single island endemics and many are restricted to one volcano within each island. Of the aster-feeding

Philodoria species included in our study, P. dubauticola, P. naenaeiella, and P. wilkesiella are single volcano endemics. The high level of endemism and the diverse,

yet host specific nature of Philodoria suggests that additional undescribed Philodoria species could be mining many Hawaiian asters.

The well-known Hawaiian silversword alliance includes approximately fifty aster taxa in three genera: Argyroxiphium, Dubautia, and Wilkesia A. Gray. The true silverswords and greenswords, Argyroxiphium, are some of the most highly protected

Hawaiian plants and include four extant and one extinct species (Wagner et al. 1999);

A. grayanum is the host of P. wilkesiella (Swezey 1940). While it is possible that the other Argyroxiphium species may serve as hosts for Philodoria, A. grayanum is the only extant member of this genus that has glabrous leaves. It remains to be seen whether the other Argyroxiphium species, which have dense pubescence on the leaf surfaces, are mined by Philodoria.

The Hawaiian endemic plant genus Dubautia contains approximately 23 endemic species (Carr 1985). Only three species (Kaua‘i’s endangered D. latifolia, and the widespread species D. laxa and D. plantaginea) are known to serve as Philodoria hosts

(Zimmerman 1978). We predict that our knowledge of Philodoria species that feed on

Dubautia has been limited by inadequate sampling of rare plant species in this genus. In the same way, the Kaua‘i endemic greensword genus Wilkesia could feasibly harbor an undescribed Philodoria species that has been overlooked by field surveys. Indeed, recent field observations have noted signs of internal feeding on leaves of Wilkesia gymnoxiphium (N. Tangalin pers. comm.), however, it has not yet been confirmed whether this damage is caused by Philodoria. A closer examination of Argyroxiphium,

Dubautia, and Wilkesia, and local populations on different islands, may yield additional new Philodoria species.

There are several Hawaiian aster genera that are not part of the silversword alliance that are absent from the Philodoria host plant record or have surprisingly few

Philodoria species known to feed from them. For example, Bidens L. and

Tetramolopium Nees are diverse Hawaiian aster radiations that lack Philodoria feeding records. The closely related genera Melanthera Rohr and Lipochaeta DC., which together comprise 16 widely distributed endemic Hawaiian plant species has yielded only two Philodoria feeding records (P. lipochaetaella and P. sciallactis). Many

Melanthera and Lipochaeta species are known from lowland Hawaiian ecosystems and nearly half of these species have become alarmingly rare (Chau unpubl. data).

Philodoria sciallactis mines leaves of M. integrifolia only at Ka‘ena Point on O‘ahu

(Zimmerman 1978). Ka‘ena Point is now a protected area and contains some of the only remaining intact coastal ecosystem where M. integrifolia exists naturally on O‘ahu

(Department of Land and Natural Resources, State of Hawai‘i 2007). Based on surveys conducted during the present study and collection localities listed in Swezey (1954) and

Zimmerman (1978), it is likely that P. sciallactis persists only within the small confines of this conservation land. With such a narrow geographic and host plant range, the monophagous P. sciallactis is perhaps the most threatened species in the genus, and may require immediate and urgent conservation prioritization.

Another aster genus that is likely to harbor Philodoria is Hesperomannia A. Gray, one of Hawai‘i’s most critically endangered lineages. This plant genus is comprised of four species, all of which are federally listed as endangered (Morden and Harbin 2013;

U.S. Fish and Wildlife Service 2015). While Swezey (1940) noted that he reared a moth similar to P. naenaeiella (CJ-142, Clade A, Figure 1-2) from H. swezeyi on Oah‘u, he

did not formally list this plant species as a host in his later synthesis of Hawaiian insect- plant relationships (Swezey 1954). Upon examination of Swezey’s P. naenaeiella samples, Zimmerman (1978) emphasized that Hesperomannia requires further investigation as a host plant of Philodoria. A recent study of dried Hawaiian

Hesperomannia leaves from the Bernice P. Bishop Herbarium revealed that H. arborescens leaves collected on Lāna‘i in 1929 were mined by an unknown Philodoria species (Johns et al. 2014). The population of Hesperomannia on Lāna‘i, however, is extirpated (Warren Wagner, Herbst, and Sohmer 1990; Morden and Harbin 2013). It is unclear whether additional Hesperomannia species serve as host plants of Philodoria, but recent field observations revealed signs of endophytophagous insect feeding on

Kaua‘i’s H. lydgatei and Maui’s H. arborescens (N. Tangalin pers. comm., K. M.

Bustamente pers. comm.).

Table 1-3. Host plant genera mined by Philodoria moths and the conservation status of their members. Number of native Number of Number of taxa in taxa for given native taxa in host plant genus plant genus Host plant genus Host plant family plant genus threatened or (number of native mined by endangered in taxa possibly Philodoria1 Hawai‘i3 extinct)2 Argyroxiphium Asteraceae 1 6(1) 3 Lipochaeta Asteraceae 2 16(4) 7 Xanthiumb Asteraceae 1 1 0 Clermontia Campanulaceae 1 31 8 Abutilon Malvaceae 2 4 3 Hibiscus Malvaceae 3 13 4 Sida Malvaceae 4 1 0 Myoporum Scrophulariaceae 1 2 0 Myrsine Primulaceae 1 19 5 Metrosideros Myrtaceae 1 13 0 Pisonia Nyctaginaceae 1 5 0 Pittosporum Pittosporaceae 1 11 3 Lysimachia Primulaceae 2 14(1) 8

Table 1-3. Continued Number of native Number of Number of taxa in taxa for given native taxa in host plant genus plant genus Host plant genus Host plant family plant genus threatened or (number of native mined by endangered in taxa possibly Philodoria1 Hawai‘i3 extinct)2 Neraudia Urticaceae 1 6 3 Pipturus Urticaceae 1 4 0 Touchardia Urticaceae 1 1 0 Urera Urticaceae 2 2 1 Abutilon Malvaceae 2 4 3 Hibiscus Malvaceae 3 13 4 Sida Malvaceae 4 1 0 Myoporum Scrophulariaceae 1 2 0 Myrsine Primulaceae 1 19 5 Metrosideros Myrtaceae 1 13 0 Pisonia Nyctaginaceae 1 5 0 Pittosporum Pittosporaceae 1 11 3 Lysimachia Primulaceae 2 14(1) 8 Neraudia Urticaceae 1 6 3 Pipturus Urticaceae 1 4 0 Touchardia Urticaceae 1 1 0 Urera Urticaceae 2 2 1 Notes: 1Zimmerman 1978, 2Wagner 1999, 3U. S Fish and Wildlife Service 2015, aincluding plant species currently assigned to Melanthera (see Chau unpubl. data), bnon-native

Surveys of other endangered Hawaiian plants also provide evidence for

Philodoria host plant associations. Remya Hillebr. ex Benth. is a genus of aster with three described species, all of which are endangered (U. S. Fish and Wildlife Service

2015). Hibiscadelphus Rock (Malvales: Malvaceae) includes eight species, six of which are extinct in the wild and the two of which are endangered (Oppenheimer et al. 2014;

U.S. Fish and Wildlife Service 2015). Herbarium leaves of Remya mauiensis and

Hibiscadelphus distans, both belonging to plant families known to be Philodoria hosts, show signs of insect damage that resemble leaf mining (C. A. J., unpubl. notes). Such

evidence, even though preliminary, warrants immediate further investigation into the host range of Philodoria, as many of these plants are critically endangered. Given the small body size of Philodoria, its preference for a diversity of host plants, the challenges of sampling from the typically remote locations in which these plants are found, and the evidence of larval mining on rare plants, it seems likely that previous sampling efforts may have failed to record Philodoria species that occur on uncommon plant species.

Because 13 of the 21 plant genera mined by Philodoria contain threatened or endangered species (Table 1-3), it is important that field surveys by research entomologists be encouraged in Hawai‘i to further elucidate host ranges of Philodoria species (Medeiros et al. 2013).

CHAPTER 2 EVIDENCE OF AN UNDESCRIBED, EXTINCT Philodoria SPECIES (LEPIDOPTERA: GRACILLARIIDAE) FROM HAWAIIAN Hesperomannia HERBARIUM SPECIMENS

The Hawaiian endemic leaf-mining moth genus Philodoria Walsingham 1907

(Lepidoptera: Gracillariidae) is composed of 30 described species (Zimmerman 1978).

Most Philodoria species are monophagous, but the genus as a whole is known to feed within the leaf tissue of 12 families of endemic Hawaiian host plants (Swezey 1954;

Zimmerman 1978). Several members of the genus are closely associated with endemic plant lineages, including the silversword alliance and the Hawaiian lobelioids, but their complete host plant range and classification remain unclear. Many of Philodoria’s known host plant genera included threatened or endangered species (IUCN 2013).

Considering the group’s extensive host plant range and the host specificity of each species, it is likely that there are undescribed Philodoria that specialize on rare members of recorded host plant genera. Such interactions would indicate a need for conservation of these native moths. Despite their unique life history and potential need for conservation, Philodoria has received little scientific attention since its description over a century ago.

Because of their extreme host specificity, leaf miners provide an excellent opportunity to study plant-insect interactions (Opler 1974). Leaf mine characteristics such as shape, length, frass quantity, and blotch formation are often distinctive at the genus or species level. Leaf mines preserved in fossils and herbarium specimens are useful media for capturing historical leaf miner activity (Crane and Jarzembowski 1980).

Recently, herbarium specimens were used to infer the historical distribution and interaction between a leaf-mining moth and its host plant (Lees et al. 2011). In the present report, we conducted a preliminary analysis of native Philodoria leaf mines

preserved in herbarium specimens stored at the Bernice Pauahi Bishop Museum’s

Herbarium Pacificum (BISH). The survey involved visual inspection of specimens of threatened or endangered species in five genera within known host plant families:

Neraudia Gaud. (Rosales: Urticaceae), Urera Gaud. (Rosales: Urticaceae),

Remya Hillebr. ex Benth. (Asterales: Asteraceae), Argyroxiphium DC (Asterales:

Asteraceae), and Hesperomannia A. Gray (Asterales: Asteraceae).

Our examination of Hesperomannia herbarium specimens revealed leaf mines, pupae, and pupal cases on the adaxial leaf surface of Hesperomannia arborescens collected from Lanai in 1929 (Figure 2-1). The Lanai population of H. arborescens is believed to be extinct (Wagner et al. 1990; Morden and Harbin 2013). Currently, only one species of Philodoria is known from Lanai, P. splendida Walsingham 1907, which mines the leaves of Metrosideros polymorpha (Myrtales: Myrtaceae), and is highly unlikely to occur on Hesperomannia. Ten described and undescribed Philodoria species are known to feed on Hawaiian Asteraceae. There is only one record from Kauai of a

Hesperomannia miner, but no adults or larvae were ever collected (Swezey 1940).

Furthermore, nine of these Philodoria species associated with asters are single-island endemics. This suggests that the mines on Hesperomannia herbarium specimens were caused by an undescribed, extinct species of Philodoria from the island of Lanai.

Figure 2-1. Philodoria pupal tents on the adaxial surface of Hesperomannia arborescens leaves from Lanai (BISH1022034).

Recent field observations by the second and third authors of the remaining wild populations of Hesperomannia species on Kauai and Maui suggest additional instances of unrecorded Philodoria species. These observations likely represent undescribed species endemic to their respective island as well. Since all Hesperomannia species are critically endangered (IUCN 2013) and the Philodoria species that feed on them are

thought to be specialists, it is crucial that we continue to gather basic biological data from these endemic leaf miners. Building on these basic life history data, we plan to study the population dynamics and evolution of Philodoria in order to prioritize conservation efforts for both these moths and their hosts.

CHAPTER 3 CONSERVATION MEDIA AS A TOOL TO RAISE AWARENESS FOR Philodoria MOTHS AND THEIR CONSERVATION

Background

Insects constitute approximately 70% of Hawaii’s terrestrial and marine animal biodiversity and the vast majority of these species are endemic to the Islands (Eldredge and Evenhuis 2002). Compared to their vertebrate counterparts, Hawaiian insects are generally poorly-known and have received relatively less scientific and conservation attention to date. The threats to native Hawaiian insects are diverse and include predation by alien introductions such as chameleons (Kraus and Preston 2012), ants, and parasitoids (Rubinoff and Jose 2010), as well as habitat loss driven by invasive plant species. The latter likely has an enormous and undocumented impact on native invertebrate species, especially ecological specialists like host-specific insects (Asquith

1995).

Some insects have been included in Hawaiian conservation initiatives since the

1970s (e.g. Gagné 1988), but new species are continually described and there is little long-term support for insect conservation. To bolster support for conservation initiatives geared towards Hawaiian insects, research entomologists have been urged to communicate more with both the public and resource managers (Medeiros et al. 2013).

Indeed, reconnecting broader society to nature is a critical step in gaining public support for biodiversity conservation (Pyle 1978; Miller 2005). Direct (nature settings largely without human intervention or control) and indirect (e.g. zoos, parks) nature experiences tend to increase interest and affection for nature across human populations (Townsend and Weerasuriya 2010). With these benefits in mind, the majority of Hawaiian conservation initiatives organize regular volunteer and community engagement

programs to introduce members of the local public to the beauty, uniqueness, and plight of Hawaiian ecosystems and organisms. Although direct experiences of this nature are valuable, they are also expensive in that they require conservation professionals to shift limited resources from fieldwork to event coordination and outreach.

Further, while direct and indirect experiences of nature are optimal, they are difficult or impossible for populations with limited access to nature. In those cases, vicarious experiences have been effective (Kahn and Kellert 2002; Kahn et al. 2009).

Nature video as a form of vicarious experience has been especially successful in reconnecting people with nature in settings where access to the natural world is restricted (Nadkarni et al. 2017).

Multimedia productions have been used to generate public awareness and appreciation of insects in Hawaii (Howarth and Mull 1992; Howarth and Gangé 2012).

Because many Hawaiian insects are confined to access-limited highland areas

(Zimmerman 1970) and ecosystems that are highly sensitive to human foot traffic, direct experience of Hawaiian nature is difficult for members of the public. In this case, vicarious experiences can be used to supplement community outreach programs that raise local interest in insects and conservation.

Hawaii has a history of positively influencing global attention to insect conservation (Howarth and Gangé 2012), mostly in the form of an increased awareness of the vulnerability of insular arthropod faunas. In order to increase local awareness for the little-known moth genus Philodoria, as well as global awareness for insect conservation in general, I produced a short film (Object 3-1) on the life history and conservation status of this endemic fauna.

Object 3-1. LEAF MINERS short film (.mp4 file 1253 MB)

Two important characteristics drove the film: awe for native Hawaiian biodiversity and culture, and awe for underappreciated organisms that are largely unknown to the

Hawaiian populace. Specific objectives were to creatively communicate findings from scientific research on Philodoria and to emphasize the need for conservation work via a local perspective with moral underpinnings, which is effective for communicating complex science messages (Feinberg and Willer 2013).

The target audiences were native Hawaiians, specifically the youth, and populations that already have an interest in science. To maximize local reach, the short film was submitted to the Hawaii International Film Festival (HIFF) and was an Official

Selection in 2016. Footage from the film was also featured in social media products disseminated by media outlets with global visibility such as National Geographic.

Engagement

• HIFF attendance: 492 • Vimeo views: 3,009 (as of 9/28/17) • National Geographic YouTube views: 36,210 (as of 9/28/17)

Future Considerations

Conservation media and related products are not meant to be a replacements of direct nature experiences in Hawaii, but rather complements to those efforts. To increase effectiveness, future investigations into the utility of professionally produced conservation multimedia in Hawaii should include discrete measures of public awareness of specific topics and messages before and after the experience.

It is likely that the quality of storytelling and production value influence the efficacy of science media as a means of vicarious experience. Conservation

organizations should seek collaboration with professional communicators to package messages successfully. Due to the amount of communication that needs to be done, it could be worthwhile for conservation organizations to regularly partner with or employ producers and communicators.

CHAPTER 4 EVOLUTIONARY HISTORY OF Philodoria

Introduction

Remote volcanic islands and their biota have profoundly shaped our understanding of evolutionary processes. Because of their extreme isolation, colonization events toward these islands are rare, and in situ speciation by successful colonists is presumed to be a major source of contemporary biodiversity. Therefore, the resulting insular biota is often highly endemic and provides unique biological systems within which to study evolution. As some of the most remote landmasses on the planet, the Hawaiian Islands host a highly endemic and well-studied insular biodiversity.

Hawaiian Geology

Hawaii has a unique and dynamic geological history. Island landmass forms while situated above the Hawaiian hotspot, a stationary magma plume that penetrates the northwesterly moving Pacific plate. Over time, individual islands that became subaerial proceed away from the hotspot following the plate’s movement, resulting in a linear chain of islands that increase with age the further they are from the plume

(Clague 1996). Islands that have moved off the hotspot are no longer building and are subject to erosional and subsidence forces, eventually reducing them to atolls and seamounts.

Three segments of the Hawaiian chain mark important geologic periods of biological relevance. The first and youngest segment, the current main “high” Hawaiian

Islands, is composed mostly of large islands with heights that presently exceed one kilometer and are the closest to the hotspot (Price and Clague 2002). These high islands extend from Hawaii Island, presently the largest and still forming over the

hotspot, to Kauai Island that is about 4.7 million years ago (Ma). The second and immediately older segment is composed of rocky pinnacles and atolls that make up the

Northwest Hawaiian Islands (NWHI), extending from Nihoa (7.3 Ma) to Kure atoll (29.8

Ma). Beyond the NWHI are sunken landmasses that mark the third segment, the

Emperor Seamounts. It is presumed that these three segments delineate geologic periods that each have shaped ancient and contemporary biodiversity on the Islands.

As islands move off the hotspot, increase in age, and diminish in size, available habitat dwindles, forcing biota to disperse or perish. Yet while the formation of islands may provide new habitat for taxa on shrinking, older islands, the productivity of the

Hawaiian hotspot has been inconsistent, with periods of reduced volcanic activity where few or no islands existed above the ocean surface. Between 33 Ma and 29 Ma, no islands were subaerial, resulting in the complete extinction of local biota, and necessitating de novo colonization of the Islands via long-distance dispersal (Clague et al. 2010). Following this period, the archipelago consisted only of small islands < 1000 m elevation, until the formation of Lisianski and Laysan at ~23 Ma. A second period of reduced volcanic activity between the formation of Nihoa at 7.3 Ma and Kauai at ~ 5 Ma led to an archipelago comprised again only of small, distantly spaced islands. This second period is of reduced activity is thought to have been an additional barrier limiting dispersal between the NWHI and the main Hawaiian Islands (Price and Clague 2002).

Because the extent to which these barriers affected the formation of present-day

Hawaiian biota remains uncertain, discovering the origin and evolutionary trajectory of the Hawaiian biota is of primary interest.

Origin of the Hawaiian Biota

Early divergence time estimations for some endemic Hawaiian lineages suggest that reduced volcanic activity between 8 Ma and 5 Ma led to an archipelago comprised of distantly-spaced, low islands, causing a substantial dispersal impediment for many

Hawaiian taxa, and that much of the contemporary Hawaiian biota colonized the archipelago after the formation of Kauai ~ 5 Ma. A review of published divergence time estimates for Hawaiian plant and animal lineages found that 12 out of 15 Hawaiian lineages colonized the Islands within the last five million years (Myr), after the current high islands began to form (Price and Clague 2002).

Some early studies focusing on dating the origin of Hawaiian lineages have inferred pre-Kauai colonization times, but the use of early molecular clock methods confounds interpretation of the results (e.g., Russo et al. 1995; Givnish et al. 2009).

Several more recent age estimates of Hawaiian arthropods, however, yield crown ages that predate Kauai, suggesting that the NWHI are a more significant colonizing source for the main high islands than was previously thought. These include the Hawaiian

Ptycta bark lice (~7 Ma, Bess et al. 2014), Megalagrion damselflies (~ 9 Ma, Jordan et al. 2003), Idiomyia and Scaptomyza fruit flies (~ 10–13 Ma, Lapoint et al. 2013; Katoh et al. 2017), the fancy-case caterpillar genus Hyposmocoma (~ 15 Ma, Haines et al. 2014), and Hawaiian Limnoxenus water beetles (~ 20 Ma, Toussaint and Short 2017). These estimates suggest that, at least for Hawaiian arthropods, the NWHI were a significant source of colonists to the main Hawaiian Islands.

Calibration Strategy

Conversion of estimated divergence times from relative to absolute is most reliably accomplished with well-placed fossil calibrations on internal nodes in a

phylogeny. The fossil record for soft-bodied organisms, however, is consistently poor

(Kidwell and Holland 2002), making divergence time estimation for the majority of those taxa problematic.

Previous evolutionary studies of endemic Hawaiian lineages have circumvented the lack of a fossil record by calibrating internal nodes of a phylogeny based on the age of biogeographic events, such as the formation of islands (e.g., Kawahara and Rubinoff

2013; Goodman et al. 2014; Haines et al. 2014; but see e.g., Toussaint & Short 2017 for a fossil-based analysis of divergence times). Although widely-used, calibrations based on biogeographic events are contentious because they assume that the age of the event is known with certainty and that the event measurably influenced the populations or species in consideration (Ho et al. 2015). In the specific case of evolution on volcanic islands, this calibration strategy assumes that divergence between taxa occurred near, or close to, the time of island formation, which may not necessarily be the case (Heads

2011; Mello and Schrago 2012).

Studies of Hawaiian taxa that use biogeographic events as calibrations to constrain internal node divergence times frequently rely on endemic clades that exhibit diversification patterns consistent with the “progression rule” (e.g. Haines et al. 2014).

The progression rule, which is most useful as a null hypothesis for investigating insular diversification patterns, posits that older taxa diversify on islands and colonize younger islands in the archipelago as they form, diverging in the process (Hennig 1966).

Typically, a calibration based on island age is only to be used for taxa that exhibit a very clear pattern of diversification reflecting the progression rule.

Despite the clear limitations, calibration strategies relying upon island ages remain widely-used in studies focusing on the origin and evolution of the Hawaiian biota. Yet, to our knowledge, no studies have examined the extent to which calibration by island ages reflect other less subjective calibration methods, namely secondary calibrations derived from available fossil-based divergence times. In the present study, we use the poorly-known endemic Hawaiian moth genus Philodoria to compare constraints using island ages versus secondary calibration on divergence time estimation for Hawaiian taxa.

Philodoria

The endemic Hawaiian leaf-mining moth genus Philodoria Walsingham 1907

(Lepidoptera, Gracillariidae) comprises 44 species, all of which feed on endemic

Hawaiian plants (Swezey 1954, Zimmerman 1978, Kobayashi et al. in prep.). The genus has an impressive host range, feeding on Hawaiian plant from six families from as many plant orders. Each Philodoria species is for the most part monophagous, feeding on the leaves of only one or very few closely related species of plants. Their host-plant associations include iconic Hawaiian taxa such as the dominant Hawaiian forest tree

Metrosideros polymorpha, member of the silversword alliance, and several of Hawaii’s most threatened and endangered plant species (Johns et al. 2016, Kobayashi et al. in prep). Philodoria are also narrowly endemic, with approximately 75% of species in the genus restricted to a single island (Zimmerman 1978, Kobayashi et al. in prep).

Despite their intriguing host associations and reliance on threatened Hawaiian plants, little is known about the evolutionary history of these moths in the Hawaiian

Islands. Estimating divergence times among species in this genus would allow studying the evolution of host-plant associations in a biogeographical framework. Given its broad

host range, and restricted geographic ranges of individual species in the genus,

Philodoria is an ideal candidate for this type of study.

In the present study, we examine the effects of calibration strategy on divergence time estimation for Philodoria. We focus on two currently common molecular dating methods for organisms that lack substantial fossil records, using a poorly understood

Hawaiian taxon that itself is also in need of conservation work. Our results estimate the crown age of Philodoria to determine whether these moths colonized the Hawaiian

Islands before or after the formation of Kauai. We also estimate the time of origin for the key host plant feeding groups.

Previous phylogenetic estimates for Philodoria were well-supported, but relied on low taxonomic and molecular sampling (Johns et al. 2016). In the present study, we generate a robust phylogenomic hypothesis for Philodoria using anchored hybrid enrichment (AHE) and transcriptome sequencing data, making this study the first analysis incorporating next-generation sequencing data for any endemic Hawaiian lineage. Not only is a dataset of this size unprecedented in Hawaiian studies, but our work provides a critical taxonomic framework for future conservation work in the genus.

We also reconstruct the historical biogeography and ancestral host plant associations for the genus. The results provide a fascinating and unique perspective on the evolutionary history of the group and their host plants in Hawaii.

Materials and Methods

Taxon Sampling

Collections of Philodoria were made between 2013 and 2016 on the Hawaiian

Islands of Kauai, Oahu, Molokai, Maui, Lanai, and Hawaii (Figure 4-1). Our collecting efforts targeted historical collection localities and previously recorded Philodoria host

plant species (Zimmerman 1978). To best gauge the true diversity and host-plant range of Philodoria, we also visited localities that were previously not surveyed for Philodoria, and collected from Hawaiian plants that are close relatives of known Philodoria host- plants.

Figure 4-1. Map of collecting sites visited in the course of this study. Map produced in Google Earth Pro.

To collect host-plant association data with confidence, we restricted our sampling to Philodoria larvae that were actively feeding on host plant leaves, these larvae were reared to adult moths for morphological and molecular analysis. Host plant leaves with active Philodoria larval activity were collected and stored in cool, dry conditions in plastic containers for rearing. Successfully reared moths were stored either in 96% ethanol, RNAlater, or flash-frozen, and then transferred and kept at -80C. Moth larvae

and leaves from which they were reared were kept as vouchers and are deposited at the McGuire Center for Lepidoptera and Biodiversity (MGCL), Florida Museum of

Natural History, Gainesville, Florida, USA. Parasitoids reared from these collections are also stored at MGCL. Identification of successfully reared Philodoria adult moths were made based on morphological (wing pattern and genitalia) comparisons of the collected specimens to type material stored at the Bishop Museum, Honolulu, HI, USA (BPBM), the Natural History Museum, London, UK (NHM), and the Smithsonian Institution,

National Museum of Natural History, Washington, D.C., USA (NMNH). Historical larval host-plant and locality data from Swezey (1954) and Zimmerman (1978) were also used to aid moth identification.

In total, 673 Philodoria adults were reared from plants collected at 42 localities across the Hawaiian Islands. Our collecting effort allowed to sample 26 of the 44 described Philodoria species (Table 4-1), as well as seven new, undescribed Philodoria species (Zimmerman 1978; Kobayashi et al. in prep.). Adult Philodoria were reared from

Asteraceae, Ebenaceae, Malvaceae, Myrtaceae, Primulaceae, Urticaceae – all the known plant families these moths feed from (Kobayashi et al. in prep). We included sequences from one representative individual per host-plant species per collection locality. While it would be best to include more than one specimen per species, we chose an exemplar taxon approach to maximize the collection localities and host plants represented in this study. Plutella xylostella (Plutellidae) and six non-Philodoria gracillariid species were included as outgroups (Figure 4-2) based on their close relationship to Philodoria (Kawahara et al. 2017). Plutellidae is closely related to

Gracillariidae, as both are in the superfamily Yponomeutoidea (Sohn et al. 2013). Our final dataset included 40 taxa.

Table 4-1. Philodoria species sequenced in this study, their host plants, and localities from which they were sampled. Date Species Plant Name Plant Family Island SequenceID Collected Latitude Longitude Hibiscus waimaeae subsp. P. aloaloiella hannerae Malvaceae Kauai AHE_43 7/9/15 - - P. auromagnifica Myrsine lessertiana Primulaceae Molokai AHE_23 1/14/14 21.118379 -156.904866 P. basalis Metrosideros polymorpha Myrtaceae Big Island AHE_41 5/30/15 19.481328 -155.279352 P. diospyrosella Diospyros sandwicensis Ebenaceae Maui AHE_3 1/4/14 20.615852 -156.29533 P. duocoremata Pipturus albidus Urticaceae Maui AHE_26 8/9/14 - - P. epibathra Dubautia carrii Asteraceae Molokai AHE_31 7/26/15 21.135447 -156.850606 P. floscula Pipturus albidus Urticaceae Big Island AHE_5 10/24/14 20.052309 -155.639782 P. hauicola Hibiscus tilaceus Malvaceae Maui TRANS.CJ-257 4/29/13 20.881126 -156.546767 P. hesperomanniella Hesperomannia arborescens Asteraceae Maui AHE_6 8/13/14 20.904463 -156.576171 P. hibiscella Hibiscus arnottianus Malvaceae Oahu AHE_14 7/27/14 21.362738 -157.791083 P. kauaiensis Wilkesia gymnoxiphium Asteraceae Kauai AHE_48 7/9/15 22.051957 -159.658935 P. keahii Remya mauiensis Asteraceae Maui AHE_62 8/9/14 20.873753 -156.618304 P. lysimachiella Lysimachia hillebrandii Primulaceae Oahu AHE_35 3/28/16 21.501325 -158.16538 P. marginestrigata Sida fallax Malvaceae Molokai AHE_60 12/30/13 21.200215 -157.208847 P. molokaiensis Lysimachia maxima Primulaceae Molokai AHE_52 7/26/15 - - P. neraudicola Pipturus albidus Urticaceae Maui AHE_11 8/9/14 20.882775 -156.574032 P. platyphylliella Dubautia platyphylla Asteraceae Maui AHE_8 8/11/14 20.681468 -156.231976 P. sciallactis Melanthera kamolensis Asteraceae Maui AHE_4 1/6/14 20.619556 -156.29476 P. serratus Dubautia knudsenii Asteraceae Kauai AHE_37 8/9/14 22.071581 -159.495277 P. splendida Metrosideros polymorpha Myrtaceae Oahu AHE_55 1/16/14 21.353205 -157.78829 P. succedanea Myrsine lessertiana Primulaceae Big Island AHE_66 4/25/16 - - P. touchardiella Touchardia latifolia Urticaceae Maui AHE_13 8/5/14 - - P. ureraella Urera glabra Urticaceae Maui AHE_7 8/12/14 20.882775 -156.574032 P. urerana Urera glabra Urticaceae Oahu AHE_17 7/27/14 - - P. wilkesiella Argyroxiphium grayanum Asteraceae Maui AHE_1 4/10/13 20.910519 -156.592075

Table 4-1. Continued Date Species Plant Name Plant Family Island SequenceID Collected Latitude Longitude P. zimmermani Pipturus sp. Urticaceae Big Island AHE_18 5/21/15 19.365255 -155.216135 P. sp. 1 Dubautia menziesii Asteraceae Maui AHE_12 8/9/14 20.662853 -156.332104 Hesperomannia P. sp. 2 lydgatei Asteraceae Kauai AHE_16 6/18/13 - - P. sp. 3 Touchardia latifolia Urticaceae Kauai AHE_21 7/24/15 22.071581 -159.495277 Diospyros P. sp. 4 sandwicensis Ebenaceae Kauai AHE_44 6/26/15 22.207121 -159.600315 P. sp. 5 Dubautia sp. Asteraceae Oahu AHE_49 4/23/13 21.353391 -157.788744 P. sp. 6 Melanthera integrifolia Asteraceae Oahu AHE_56 7/6/14 21.572609 -158.275253 Lysimachia P. sp. 7 kalalauensis Primulaceae Kauai AHE_20 7/2/15 22.150762 -159.636946

Sample Preparation, Sequencing, Data Processing, and Dataset Construction

We used anchored hybrid enrichment (AHE; Lemmon et al. 2012) to infer a robust phylogeny of Philodoria. We used the across-Lepidoptera AHE (Lep 1) probe set

(Breinholt et al. 2017) to capture 855 loci from 33 Philodoria samples. For one

Philodoria sample (P. hauicola, CJ-257, Table 4-1), the sequence data were initially collected as a transcriptome and trimmed to the 855 AHE loci. This species was sampled as a transcriptome because it was included in the original design of the Lep1 probe set. The RNA extract of P. hauicola was isolated following the Illumina TruSeq

RNA Low Sample Protocol (Illumina 2014) and then sequenced on a single lane of paired-end, 150bp Illumina HiSeq 2000.

For all other samples, DNA was extracted from ethanol-stored Philodoria tissue using the OmniPrep Genomic DNA Extraction Kit (G-Bioscience: Catalog #786-136; St.

Louis, MO, USA). DNA extracts were processed at Florida State University’s Center for

Anchored Phylogenomics (www.anchoredphylogeny.com), Tallahassee, FL, USA and at

RAPiD Genomics (www.rapid-genomics.com), Gainesville, FL, USA. Library preparation followed protocols detailed in Lemmon et al. (2012) and Breinholt et al. (2017). Libraries were sequenced on a single lane of paired-end, 150bp Illumina HiSeq 2500.

Transcriptome sequences for P. hauicola (CJ-257) and outgroups were processed following data cleaning and assembly methods of Breinholt and Kawahara

(2013). The eight-step custom pipeline of Breinholt et al. (2017) was used to process raw AHE sequence data. For both sequence data types, we predicted orthologous genes using the Bombyx mori genome (Xia et al. 2004) as reference, and then concatenated the two data types prior to step 5 in the Breinholt et al. (2017) pipeline.

We conservatively chose to use ortholog locus alignments that contained sequences for at least 70% of the sampled taxa, resulting in a 507-locus, concatenated starting dataset for phylogenetic analysis. Each individual locus alignment was visually inspected in AliView v1.17.1 (Larsson 2014) and edited to ensure the correct reading frame without internal stop codons, and to remove poorly-aligned regions.

To assess whether rapidly evolving genomic regions can resolve species-level relationships, we constructed two versions of the 507-locus dataset, one without flanking regions (Dataset 1), and another with them (Dataset 2). Dataset 1 included the transcriptome sequence and AHE sequences cut to the probe region. Dataset 2 included the transcriptome sequences cut to the probe region and the full length AHE loci with sequence data flanking the probe region. The Lep1 AHE kit was designed to fit the capture probes within exonic regions, resulting in flanking data on either side of the probe region that consists almost entirely of intron sequences (Breinholt et al. 2017).

We aligned these “flanks” with MAFFT v 7.245 (Katoh and Standley 2013), using the commands “-allowshift -unalignlevel 0.8 -reorder -leavegappyregion.” Unalignable regions were removed using custom Python scripts (alignment_DE_trim.py and flank_dropper.py) developed by Breinholt et al. (2017) that trim data based on density (# sequences with data/ total # sequences in alignment) and nucleotide entropy (Xia et al.

2003). We chose to remove alignment columns with <60% density and >1.5 (nearly random) nucleotide entropy to target genetic variation at the species-level (see Dataset

4 of Breinholt et al. 2017). Alignment statistics and dataset completeness were calculated using AliStat v 1.3 (Misof et al. 2014). Final concatenated alignments for

Dataset 1 (119,323 bp) and Dataset 2 (258,995 bp) datasets are deposited in the Dryad

Digital Repository (datadryad.org).

Phylogenomic Analyses

Phylogenomic analyses were performed using maximum likelihood (ML) and a coalescent-based method that estimates a species tree from multiple gene trees

(ASTRAL-II). For ML analyses, we used PartitionFinder2 (Lanfear et al. 2017) to determine the optimal partitioning scheme of both Dataset 1 and Dataset 2 using the rcluster search algorithm (Lanfear et al. 2014). For Dataset 2, all flanking data were assigned independently to one partition because these sites were not necessarily contiguous. Model selection was performed in IQ-TREE v 1.4.2 (Nguyen et al. 2015) using the command “-m TESTNEW” on the best partition schemes determined through

PartitionFinder2. Partition-specific rates were allowed in IQ-TREE using the “-spp” option. We assessed support for all resulting ML phylogenetic trees using 100 nonparametric bootstrap (BS) replicates as well as 1000 ultrafast bootstrap (UFBoot) replicates in IQ-TREE. UFBoot support values have been shown to provide a less biased approximation of clade credibility (Minh et al. 2013). We also performed a

Shimodara-Hasegawa-like approximate likelihood-ratio (SH-aLRT) test to assess internal branch support using 1000 replicates (Guindon et al. 2010).

Species-tree estimation was performed in ASTRAL-II v 4.10.12 (Mirarab et al.

2014) using the full-length loci of Dataset 2. ASTRAL-II is especially useful in accounting for gene tree discordance due to incomplete lineage sorting. Individual gene trees were estimated using the most appropriate model for nucleotide evolution for each locus in Dataset 2 and 100 rapid bootstraps in RAxML v 8.2.3 (Stamatakis 2014).

Internal node support for the ASTRAL-II species tree was calculated from 100 bootstrap replicates for each gene.

Divergence Time Estimation

Estimating the evolutionary timescale on which Philodoria diversified in the

Hawaiian Islands is a primary goal of this study. However, a series of computational and theoretical challenges emerge when estimating divergence times for a large phylogenomic dataset and when the focal taxon lacks a fossil record.

Rigorous divergence time estimates using large phylogenomic datasets are computationally intensive, and often prohibitively so. Computationally fast methods for estimating evolutionary dates exist, but their accuracy remains unclear (Lozano-

Fernandez et al. 2017). Given the proliferation of large datasets in evolutionary studies, a fast, repeatable, and accurate analytical divergence time approach is greatly needed.

We addressed this challenge by subsampling Dataset 1 (probe region only) to make it more tractable for Bayesian relaxed clock methods by calculating a Robinson-

Foulds (RF) distance (Robinson and Foulds 1981) between individual gene ML trees and the ML tree from the full-length Dataset 2. The RF metric measures topological similarity between sets of trees. Using the program HashRF (Sul and Williams 2008), we calculated the RF distance between the gene trees estimated in RAxML for the

ASTRAL-II analysis above to the maximum likelihood tree inferred in the IQ-TREE analysis of Dataset 1. Because our aim was primarily to estimate divergence times for

Philodoria, we chose to subsample our dataset to represent the phylogenetic hypothesis constructed as part of this study, rather than assemble and test random sets of genes to approximate the entirety of gene histories. The 50 loci whose gene trees were closest in

RF distance to the Dataset 1 ML tree were used as input for our divergence time estimation.

To ensure that the 50-locus subsampled dataset accurately represents Dataset

1, we ran a greedy search for the optimal partitioning scheme in PartitionFinder2. We used the best resulting partitioning scheme to conduct an ML phylogenetic tree search in IQ-TREE with 100 bootstrap replicates. The 50-locus subsampled dataset tree and the Dataset 1 tree topologies were compared using the similarity index in phylo.io

(Robinson et al. 2016).

To obtain absolute divergence times, we used two calibration strategies applied to the 50-locus subsampled dataset. The first strategy applied a secondary calibration to the node connecting the outgroup taxa Phyllocnistis and Caloptilia in our 50-locus dataset tree. We constrained that node using a uniform prior distribution of 88.5476-

115.9883 Myr based on the study of Wahlberg et al. (2013), which estimated divergence times for Lepidoptera using six fossil calibrations and included representatives of both Phyllocnistis and Caloptilia.

Our second calibration strategy relied upon a biogeographic calibration. We ran a separate divergence time analysis using the ages of Hawaiian island formation as node constraints to cross-examine the effects of the widely-used island calibration approach to our secondary calibration approach from Wahlberg et al. (2013). We calibrated four nodes in our phylogeny based on the age of formation of Oahu and the Maui Nui island group. We treated the islands that comprise Maui Nui as a group because they were connected for much of their geological history. Two nodes suggesting dispersal from

Kauai to a recently formed Oahu were calibrated using a normally distributed prior for

those nodes set to 3.0 Ma (s.d. 1.0). In addition, two nodes representing dispersals from

Oahu to a recently formed Maui Nui were also calibrated using a normally distributed prior age of 2.2 Ma (s.d. 0.73). Normally distributed priors were used to account for uncertainty in the actual time of colonization by the taxa derived from the calibrated node. Standard deviations of these priors were conservatively set to one-third the island age, to help narrow the island age most biologically-relevant for colonization. We used this approach to maintain consistency and allow comparison with the calibration methods in other recent studies of diversification in endemic Hawaiian lineages (e.g.

Haines et al. 2014).

To improve our confidence in the results of the divergence time estimation for each calibration strategy, we also tested the effects of other parameters including speciation process (Yule versus birth-death), partitioning versus no partitioning of the

50-locus subsampled dataset, and using the 50-locus ML tree versus the ML tree from

Dataset 2 as the constraint tree. We also tested the effects of constraining the maximum age of the tree root using an additional calibration from Wahlberg et al., by designating a uniform prior distribution of 88.5476-141.2642 Myr reflecting the upper age bound for the MRCA of Plutella and Gracillariidae in their study (2013).

All divergence time estimation analyses were conducted in BEAST 1.8.4

(Drummond et al. 2012) on the CIPRES Science Gateway. Each BEAST MCMC analysis used an uncorrelated relaxed lognormal clock prior and was run for 50,000,000 generations, sampling every 5000th generation. To allow comparison amongst all our divergence time analyses, we estimated the marginal likelihood (MLE) for each of the individual BEAST runs using the path/stepping-stone sampling algorithm using the

settings: number path steps = 100, chain length = 1,000,000, and log likelihood sampling = 1000. Chain convergence was assessed by effective sample size (ESS) greater than 200 for each estimated parameter in Tracer v 1.6.0. Trees were summarized in TreeAnnotator v 1.8.4 (Drummond et al. 2012) and 10% of trees from each run were removed as burn-in. We used FigTree v 1.4.2 (Rambaut and Drummond

2009) to view all trees.

Historical Biogeography and Ancestral Host Plant Reconstruction

We estimated the historical biogeography of Philodoria using BioGeography with

Bayesian (and likelihood) Evolutionary Analysis (BioGeoBEARS; Matzke 2013) implemented in the R statistical environment (R Development Core Team 2014).

BioGeoBEARS allows to infer ancestral range evolution under different models, each of which allows/disallows various combinations of biogeographic events to explain range inheritance along a lineage. In BioGeoBEARS, we estimated ancestral range using the

DEC model (Dispersal-Extinction-Cladogenesis; Ree et al. 2005; Ree and Smith 2008).

Outgroups were pruned from the ultrametric MCC tree that was inferred from the most likely (based on marginal likelihood estimate) BEAST analysis above. The distribution of each species (as assembled by Kobayashi et al. in prep) on five biogeographic areas (Nihoa, Kauai, Oahu, Maui Nui, and the island of Hawaii) was assigned to respective tips. Lanai, Molokai, and Maui, were treated as a single area

(Maui Nui) because these islands, which are presently separated, were joined as one large island during the numerous, recurrent Pleistocene glacial low stands. In all biogeographic analyses, the maximum number of areas that an ancestor could inhabit was set to five.

We performed multiple ancestral range estimations to test the impact of dispersal area constraints and two additional estimated parameters on the DEC model’s biogeographic reconstruction of Philodoria. Using the +J parameter in BioGeoBEARS, we evaluated the significance of founder event speciation on range evolution on

Philodoria. This mode of speciation is characterized by a rare dispersal event to a new area that results in instantaneous isolation, making it a theoretically important process in historical biogeography of insular taxa. Additionally, we estimated the effects of dispersal area distance by incorporating the +x parameter, which modifies the dispersal rate based on distance. Distance between the centers of each combination of two island areas were measured and then rescaled by dividing by the shortest length, so as not to influence likelihood estimation. Finally, to further assess the impact of spatial separation on ancestral range estimates, we disallowed dispersal to areas that were not adjacent.

In sum, we ran eight BioGeoBEARS analyses: (1) DEC model unconstrained, (2) DEC model +J, (3) DEC model unconstrained, dispersal to adjacent areas only, (4) DEC model +J, dispersal to adjacent areas only, (5) DEC model +x, dispersal to adjacent areas only, (6) DEC model +J +x, dispersal to adjacent areas only, (7) DEC model +x, and (8) DEC model +J +x. We compared the statistical fit of each model using the

Akaike information criterion (AIC) and the Likelihood Ratio Test (LRT).

Philodoria species are extremely host-plant specific, but the genus has a broad host range spanning six plant families. Previous work has suggested that host shifts have played a major role in gracillariid diversification (López-Vaamonde et al. 2003;

Kawakita and Kato 2009; Kawahara et al. 2017). To better understand the evolutionary role of host switching in Philodoria, we reconstructed the ancestral host plant

associations across the genus. Host plant family was coded for each species and mapped onto the ultrametric MCC tree from the most likely BEAST analysis. Host range evolution was then estimated under the maximum parsimony criterion in Mesquite v.

3.31 (Maddison and Maddison 2017). We chose to use a simple optimality criterion to avoid misspecification with more complex models.Results

Phylogenomics

The number of captured loci for the 40 taxa included in this study ranged from

169 – 621 (Figure 4-2). The final concatenated nucleotide alignment length was

119,323 bp for Dataset 1 (86% complete) and 258,995 bp for Dataset 2 (60% complete). For Dataset 2, P. hauicola and outgroup taxa that were sampled as transcriptomes accounted for most of the missing data, because these sequences lacked data that flank the probe region. PartitionFinder2 found 28 partitions for Dataset

1 and 39 partitions for Dataset 2. Models for each partition are listed in supplementary information (Table B-1 and B-2).

The ML tree estimated from Dataset 2 (full-length loci) is shown in Figure 4-3.

Results were largely congruent between datasets. Dataset 2 had higher branch support

(BS  80% and UF-Boot/SH-aLRT  95%/80%) for all but three ingroup nodes, compared to the tree from Dataset 1 which lacked support for 4 of 32 ingroup nodes.

Several nodes had moderate support (BS < 80%, UF-Boot/SH-aLRT <

95%/80%) in all analyses despite the addition of flanking data. The relationship between

(P. hesperomanniella, Philodoria sp. 2) and (P. kauaiensis, P. serratus, Philodoria sp. 5) was moderately supported across ML analyses (BS < 80%, UF-Boot/SH-aLRT <

95%/80%). Results from the ASTRAL-II analysis largely reflected those of the ML

estimations (Figure B-1). Support throughout the tree was strong, with BS  80% for all but five nodes.

Figure 4-2. ALISTAT heatmap of pairwise nucleotide completeness for the final nucleotide alignment of Dataset 1. Completeness refers to the percentage of alignment columns in which data is present between two taxa. Numbers in parentheses next to taxon names indicate number of captured loci.

Figure 4-3. Maximum likelihood tree of Philodoria estimated from Dataset 2 (full-length loci), with the support values from both datasets including Dataset 1 (probe region only) and from both bootstrapping methods included for comparison.

Figure 4-4. Violin plots of Philodoria crown age from most likely divergence time analyses. Arranged from bottom to top in order of increasing likelihood. Analyses based on only island calibrations result in younger crown age estimates than analyses incorporating fossil-based constraints. Summary of analyses values in Table 4-2.

Table 4-2. Divergence time analysis results, with ages in millions of years with 95% highest posterior density range in parentheses. Clade names reflect those indicated in Figure 4-3. Analysis names descriptors as follows: part/unpart = data partitioned or unpartitioned; 507/50 = 507 or 50 locus topology used as constraint; Wahlberg/islands = calibration strategy; Yule/BD = speciation process using Yule or birth-death prior. Analyses that did not converge are those without MLE scores and age estimates. Analysis Name MLE score Philodoria Crown Age part_50_wahlberg_BD -120020.3611 24 (18.9281-27.7147) part_50_TOTAL_yule -120343.7631 20 (17.4885-22.0707) part_507_wahlberg+treeroot_yule -120375.0442 22 (18.5486-26.8013) part_50_wahlberg+treeroot_yule -120377.4512 24 (19.6182-28.4079) part_50_wahlberg+treeroot_BD -120381.479 23 (19.3512-27.8454) part_50_islands+treeroot_BD -120389.5796 19 (16.8092-21.3164) part_50_TOTAL_BD -120390.6177 19 (17.2182-21.737) part_50_islands_yule -120391.9453 10 (5.8284-13.4216) part_50_islands+treeroot_yule -120398.4324 19 (17.0952-21.5372) part_507_islands_BD -120403.268 10 (5.8506-13.5322) part_507_wahlberg_BD -120404.0802 22 (18.1015-26.1616) part_507_wahlberg+treeroot_BD -120404.5646 22 (17.9754-26.0804) part_507_islands+treeroot_BD -120405.1282 18 (15.805-20.4024) part_507_TOTAL_BD -120405.141 18 (16.112-20.6287) part_507_islands_yule -120411.8199 10 (5.7253-13.5558) part_507_islands+treeroot_yule -120417.3065 18 (16.1705-20.7778) part_507_TOTAL_yule -120419.4154 19 (16.4631-21.0761) unpart_50_islands+treeroot_yule -120523.2929 20 (15.1275-25.0055) unpart_50_wahlberg+treeroot_yule -120571.1336 29 (21.6349-36.374) unpart_50_wahlberg+treeroot_BD -120960.0109 27 (20.0911-34.2359) unpart_50_islands+treeroot_BD -120965.2759 19 (14.6096-23.646) unpart_507_wahlberg+treeroot_BD -120970.0236 24 (17.8183-30.2286) unpart_507_islands+treeroot_BD -120973.1813 18 (13.2571-22.7947) unpart_507_wahlberg+treeroot_yule -120977.8932 26 (19.3159-32.7821) unpart_507_islands+treeroot_yule -120980.6909 19 (14.1123-23.3444) part_50_islands_BD - - part_50_wahlberg_yule - - part_507_wahlberg_yule - -

Divergence Time Estimation

The tree inferred based on the 50-locus (19,077 bp) subsampled dataset was nearly identical to the ML tree from Dataset 1, with all Philodoria host plant family feeding group relationships preserved and only minor topological differences (Figure B-

2). Nearly all BEAST analyses using the 50-locus subsampled dataset converged with effective sample size (ESS) values well above 200 (Table 4-2). The relaxed molecular clock estimates for the crown age of Philodoria from each of the divergence time analyses that used secondary calibrations from Wahlberg et al. (2013) largely agree, with overlapping credibility intervals (Figure 4-4, Figure 4-5). Philodoria crown age estimates that were based exclusively on island calibrations, however, were consistently younger than the fossil-based estimates. The analysis with the best marginal likelihood score (part_507_Wahlberg_Yule, Table 4-2) used the secondary calibration from

Wahlberg et al. (2013) with the tree root height constrained, and estimated the crown age divergence between Philodoria and its closest relatives in the early Miocene at 23

Ma (95% HPD = 18.55-26.80). The Primulaceae-feeding clade was estimated to have originated at 11 Ma (95% HPD = 8.84-12.98; Figure B-2). Our analyses suggest that

Urticaceae-feeding Philodoria evolved at 7 Ma (95% HPD = 4.65-6.58), before the formation of Kauai. The crown age of the extant Asteraceae-feeding clade is dated at 4

Ma (95% HPD = 3.46-5.03), while Kauai island was still forming over the hotspot. The most likely analysis using the biogeographic calibration based on island ages

(part_50_islands_BD, Table 4-2) dated the crown age of Philodoria at 10 Ma (95% HPD

= 5.8284-13.4216), considerably younger than the calibration strategy that used a secondary calibration from Wahlberg et al. (2013).

Historical Biogeography and Ancestral Host Plant Association

All BioGeoBEARS analyses were run using the DEC model (Table 4-3). The analyses that also added the jump dispersal parameter (+ J) and the dispersal matrix multiplier (+ x) to the DEC model outperformed analyses that did not. The preferred model, based on AIC score, was the DEC + J + x with dispersal to new islands limited to those immediately adjacent (Figure 4-6). Because we did not include time stratification in our analyses, our ancestral range estimates at deeper nodes in the tree are untenable.

Our host plant association reconstruction suggests that the ancestral Philodoria that colonized the Hawaiian Islands was polyphagous, likely feeding on plants in the families Ebenaceae, Malvaceae, and/or Primulaceae (Figure 4-7). Other major

Philodoria host associations were reconstructed to have fed on a single host family relatively early in the evolution of the genus, namely such Urticaceae (~7 Ma) and

Asteraceae-feeding Philodoria (~4 Ma).

Figure 4-5. MCC chronogram from BEAST, with nodes ages in millions of years and blue bars at nodes representing the 95% highest posterior density range.

Table 4-3. BioGeoBEARS results. Parameters = d (dispersal), e (extinction), j (jump dispersal), and x (dispersal modified by distance). LRT values compare pairs of models with nested parameters. Constraints # Model model parameters d e j x LnL AIC deltaAIC p-val (LRT) DEC unconstrained 2 0.0471 0.0167 0 0 -95.22 194.44 32.33 DEC + J unconstrained 3 0.0365 0 0.0703 0 -90.69 187.38 25.27 0.003 area adjacency DEC constraint 2 0.0691 0.0168 0 0 -86.28 176.57 14.46 area adjacency constraint + DEC + J jump dispersal 3 0.0567 0 0.058 0 -83.37 172.75 10.64 0.016 area adjacency constraint, modified by DEC + x distance (x) 3 0.1631 0.0096 0 -2.8312 -82.13 170.25 8.14 area adjacency constraint, + jump dispersal, modified by DEC + J + x distance (x) 4 0.1345 0 0.221 -2.8626 -77.06 162.11 0 0.001 no area adjacency constraint, modified by DEC + x distance (x) 3 0.1321 0.0056 0 -2.3721 -83.48 172.96 10.85 no area adjacency constraint, + jump dispersal, modified by DEC + J + x distance (x) 4 0.1393 0 0.2452 -3.2941 -77.33 162.66 0.55 0.000

Figure 4-6. Ancestral state reconstruction of Philodoria host plant associations using the MK-1 model.

Figure 4-7. Historical biogeography of Philodoria inferred by BioGeoBEARS. Areas are as follows: N = Nihoa, K = Kauai, O = Oahu, M = Maui Nui, H = Hawaii Island.

Discussion

Phylogenomics

The present study provides the first molecular phylogeny constructed from next- generation sequencing data for an endemic Hawaiian animal lineage. Our results, based on 507 loci, strongly support the monophyly of and relationships within

Philodoria, across optimality criteria and dataset types. As reported in other phylogenomic studies using anchored hybrid enrichment data, the addition of flanking data strengthened support for nodes that were weakly supported in analyses that only incorporated data from the probe region.

This study includes the largest gene and taxon sampling of Philodoria for phylogenetic analysis to date; more than 150 times the number of genes and double the number of Philodoria species included in previous work (Johns et al. 2016). Our results are consistent with the previous phylogenetic estimates of Johns et al. (2016) that were based on Sanger sequencing data, finding robust nodal support for the two major clades in the genus in addition to the majority of clades representing the major

Philodoria host plant family feeding clades. Our results also agree with recent work

(Johns et al. 2016) in rejecting the subgeneric classification of Philodoria proposed by

Zimmerman (1978). In addition, our phylogenomic analyses confidently place five recently described and seven currently undescribed Philodoria species with previously sequenced taxa that feed on plants in the same family, respectively, which provides further evidence that Philodoria relationships are mostly conserved at the host-plant family level.

Nodes that received poor support in more than one analysis, despite the inclusion of flanking data, are likely due to rapid radiations into new habitats and/or onto

new host-plants that is reflected in the genome by incomplete lineage sorting or hybridization events. Our divergence time analysis, supports this hypothesis, given that major feeding clades in Philodoria originated near the formation of Kauai, a time that is thought to have seen a proliferation of ecosystem types for diversification and potential host plants for Philodoria. We hypothesize that a rapid increase in the availability of ecosystems and host plants near the time Kauai formed may have facilitated an increase in the diversification rate in Philodoria. The ASTRAL analysis also found low support for the relationships within these clades, lending credibility to this hypothesis.

Dataset Subsampling

Devising a fast, accurate, and repeatable workflow for divergence time estimation using a large phylogenomic dataset (hundreds of loci) was a primary analytical goal in this study. On the premise that missing data have a relatively minor effect on divergence time estimates (Filipski et al. 2014; Zheng and Weins 2015), we chose to rigorously select individual genes that represented the phylogeny estimated from the

507 loci of Dataset 1, to make our phylogenomic dataset more tractable for Bayesian relaxed clock analyses. Because our foremost conceptual objective in this study was to estimate the time Philodoria colonized the Hawaiian Islands, we first generated a robust phylogenetic hypothesis and then relied on this framework to estimate divergence times in the genus. The ML tree topology of the 50-locus (19,077 bp) subsampled dataset was congruent with the tree constructed from Dataset 1, with all relationships between major

Philodoria feeding clades preserved, and with only minor discrepancies between species pairs within each feeding clade (Figure B-2).

The 50-locus dataset alignment length was ~6% of the alignment length of

Dataset 1. The small size of our 50-locus dataset allowed us to test the effect of various

parameters and configurations in iterative BEAST runs but also reach convergence

(ESS > 200) in most analyses. Future studies attempting to subsample phylogenomic datasets for divergence time analysis should seek to maximize input dataset size while maintaining the computational ability to test multiple BEAST configurations and achieve convergence. While the 50 subsampled loci adequately approximated our larger phylogenomic dataset, the optimal number of genes that are required to accurately represent a next-gen dataset will vary for other taxonomic groups. We encourage careful testing of various combinations of subsampled sequences to capture a representative subset of data for divergence time estimation. Additionally, we only recommend this approach for strongly-supported phylogenies that are appropriate given the specific evolutionary hypothesis. We suggest that future phylogenomic studies test multiple subsampling methods, including other Robinson-Foulds metric value cutoffs, and investigate the effect of random samples of gene trees to increase representation of all gene histories.

Calibration Strategy

Adaptive radiations on remote oceanic archipelagos provide ideal biological systems within which to study evolutionary processes. Nowhere else is this opportunity better demonstrated than in the Hawaiian Islands, where a significant proportion of the natural biodiversity is composed of iconic adaptive radiations, many of which have profoundly shaped our understanding of the forces that drive diversification.

Understanding the tempo of diversification in Hawaiian lineages is made possible by a stationary volcanic plume situated beneath the Pacific plate. This hotspot, in combination with the plates northwesterly direction of movement, produces a linear chain of islands that increase in age as they move away from the hotspot. A significant

proportion of studies on Hawaiian evolution have used the spatial arrangement and known ages of islands to convert the observed rates of change in molecular data to absolute time scales.

Application of biogeographic calibrations is controversial because their use assume that a particular event had a clear and measureable impact on an organism.

However, phylogeny calibration using island ages is often the only means for generating divergence times for most Hawaiian taxa, due to a fossil record that is depauperate for soft-bodied organisms. To our knowledge, no studies that have focused entirely on an endemic Hawaiian clade have compared alternative divergence time estimates due to biogeographic versus secondary calibration. Here, we compared Philodoria divergence time estimates from two independent calibration schemes, one that applied a biogeographic constraint based on island ages and another using a secondary calibration from a dated tree for all of Lepidoptera (Wahlberg et al. 2013). For our biogeographic calibration, we conservatively chose to only calibrate nodes that showed signs of islands colonization via the progression rule (as recommended by Ho et al.

2015), a pattern of diversification where one of a pair of taxa inhabits an older island and the other taxon inhabits an immediately adjacent and younger island.

Our series of fossil-based divergence time analyses returned largely congruent results for the crown age of Philodoria, as well as for the evolution of each major feeding clades in the genus (Figure 4-4, 4-5). In contrast, the analyses that used only biogeographic calibrations resulted in considerably younger divergence time estimates for Philodoria. These results show that, for this specific Hawaiian taxon, the two calibration strategies are not comparable in their ability to address the evolutionary

hypothesis at hand, and that biogeographic calibration underestimates the timing of major divergences in Philodoria.

A primary line of inquiry in Hawaiian studies is estimating the age of colonization for endemic lineages, because it allows us to better understand the source of contemporary Hawaiian biodiversity. Given the recent advent of more sophisticated methods for divergence time estimation and the availability of large phylogenomic datasets, our findings pave the way for the study of other Hawaiian lineages. This is potentially an even more interesting and pressing line of questioning in Hawaii today, given the conservation and extinction crisis in the archipelago. Moreover, the advent of new sequencing methods such as AHE makes possible the integration of samples from natural history collections.

Evolution of Philodoria and Their Host Associations

Our discussion on the evolutionary history of Philodoria is centered on the part_507_Wahlberg_yule analysis (Table 4-2, Figure 4-5) because it is based on the full phylogenomic dataset topology and was the most likely estimate given the data. Data presented here offer a unique perspective on the evolution of endemic Hawaiian taxa.

These results suggest that Philodoria, unlike the majority of dated Hawaiian lineages, originated in the Islands at 23 Ma (95% HPD: 18.55-26.80), before the formation of

Kauai but prior to the first peak period for colonization proposed by Price and Clague

(2002). They hypothesized that colonization of the Hawaiian Islands was most likely when multiple volcanic peaks exceeded 1000 m. These once-high islands received more rainfall and thus could support a greater diversity of terrestrial ecosystems at their peak heights. Based on estimated island age and maximum island size, Price and

Clague (2002) depicted two “peak periods” for putatively higher rates of colonization in

Hawaii, the first between 8-18 Ma and the second between 3 Ma and the present. Our divergence time analysis indicates that Philodoria colonized Hawaii just before the first peak period, when Laysan and Lisianski, which today are only low landmasses without forest ecosystems, were near their peak heights. These findings are in line with the hypothesis that Lisianski was perhaps the first island in the chain that the contemporary

Hawaiian biota could colonize, because prior to its existence, there was a period where no islands were subaerial, in turn forcing colonization to completely restart (Clague

2010). Our data indicate that Philodoria colonized the Hawaiian Islands shortly after dispersal to the archipelago was a possibility.

At the approximate time Philodoria colonized the archipelago, Lisianski and

Laysan were the largest landmasses in the archipelago, and exceeding 1000 m in elevation, likely harbored a diversity of forest ecosystems and prospective host plants.

While our results suggest that Philodoria inhabited the NWHI at one point in the past, it is unlikely that the Philodoria species that depend on forest-inhabiting host plants still exist in the NWHI today, due to the reduction in habitat and host availability. Biodiversity surveys on several NWHI have never recorded Philodoria (Conant et al. 1984).

Our divergence time estimates also suggest that Philodoria on older Hawaiian

Islands may have colonized younger islands within the archipelago as they formed over the hotspot during the first peak period, a pattern of diversification that is consistent with the progression rule. The origin of Clade A (Table 4-2) at 13 Ma (95% HPD: 10.81-

15.67) indicates that divergence of these taxa occurred at around the time when

Gardner, LaPerouse, and Necker were islands with at least one > 1000 m peak and theoretically supported diverse ecosystems. Clade B (10 Ma; 95% HPD: 7.69-11.61)

follows the same pattern, originating before the formation of Kauai and also within the first peak period.

Price and Clague (2002) proposed that a period of reduced volcanic activity between ~11-5 Ma, after Necker had formed but before the emergence of Kauai, has significantly shaped contemporary Hawaiian biodiversity. During this geological period, the archipelago was greatly reduced with distantly spaced volcanic peaks with elevations below 1000 m, resulting in a decline in complex terrestrial habitats. This reduction of ecosystems is thought to have been a significant colonization barrier for taxa on islands that predate Kauai to disperse to the current high islands. Philodoria survived this period of low volcanic activity and successfully dispersed to the current high islands from older, now submerged Hawaiian Islands. Our data imply that eight lineages of Philodoria dispersed from the NWHI to the current high islands.

Other studies have dated the colonization of the Islands for extant Hawaiian arthropod lineages prior to the formation of Kauai, but only one of these studies utilized external calibration points (Toussaint and Short 2017). The megadiverse Hawaiian case-building caterpillar genus Hyposmocoma is an extreme example of a pre-Kauai origin (Haines et al. 2014), with over twenty independent dispersal events from the

NWHI to the current high islands. Only one Hawaiian plant lineage (the Hawaiian lobelioids) has been found to have colonized the current high islands from the NWHI

(Givnish et al. 2009). The review of dated Hawaiian lineages by Price and Clague

(2002) led them to suggest that colonization by the present-day Hawaiian biota was more likely after the formation of the current high islands. Our results, along with recent studies that use rigorous Bayesian relaxed clock methods to date divergences of

Hawaiian taxa, reflect mounting evidence suggesting that a larger fraction of contemporary Hawaiian biodiversity originated in the NWHI than proposed by Price and

Clague (2002).

Our phylogenomic analysis included representatives of Philodoria that feed on all six plant families known to be host-plants. Previous work (Zimmerman 1978; Johns et al. 2016) indicated a higher number of Philodoria host plant family associations, but the most recent treatment of the genus (Kobayashi et al. in prep), based on extensive collecting efforts and a complete survey of museum-stored specimens and their label data, restricted the reliably documented host range to six plant families: Asteraceae,

Ebenaceae, Malvaceae, Myrtaceae, Primulaceae, and Urticaceae. Our reconstruction of ancestral host plant associations for Philodoria indicate the presence of clear host- lineages, which are robust at the plant family level. These results demonstrate that these moths may have fed on members of these plant families on former high islands

(Figure 4-5).

Three major host-plant family feeding clades (Ebenaceae, Malvaceae, and

Primulaceae), on which a significant proportion of Philodoria diversity feeds, evolved prior to dispersal to the current high islands. This suggests that these host-plant families once existed on the NWHI, whereas today they are entirely absent. Several lines of evidence support this hypothesis. Because these plant families are represented on the current high islands by members that inhabit coastal, dryland, and rock habitats, it seems within reason that relatives of these plants may have once served as host to

Philodoria on the NWHI. As confirmation, the host plant for Philodoria marginestrigata,

Sida fallax (Malvaceae), is presently found in the NWHI on both Midway atoll and Nihoa

(Wagner 1990). The pollen record also shows that this plant species was found on

Laysan (Athens et al. 2007). Sida fallax has since been extirpated from this island with the introduction of non-native herbivores (Rauzon 2001). Similarly, pollen from Hibiscus

(Malvaceae), on which at least three extant Philodoria species feed (Kobayashi et al. in prep), was also found on Laysan, but is now extirpated as well (Athens 2007). The relatively well-documented extinction of the endemic Laysan weevil Rhyncogonus bryani due to defoliation of its host plant by non-native herbivores (Conant et al. 1984;

Asquith 1995) may provide insight into why Philodoria is no longer found on that island.

The divergence times for extant Hawaiian members of Primulaceae and Ebenaceae are unknown, but their ecological affinities and the distribution of non-Hawaiian relatives do not prevent a possible colonization of the NWHI. Host of at least three extant Philodoria species (Kobayashi et al. in prep), species of the Hawaiian genus Lysimachia

(Primulaceae) are monophyletic and are known to disperse via ocean currents (Hao et al. 2004; Kono et al. 2012).

Like other pre-Kauai Hawaiian arthropod lineages (e.g. Hyposmocoma,

Limnoxenus), several Philodoria clades evolved soon after the formation of the current high islands (Figure 4-5). Cladogenesis at that time could reflect an increase in available niche space, as new habitats developed and prospective host plants emerged on Kauai. For example, the most diverse Philodoria species-group, the Asteraceae miners, originated at 4 Ma (95% HPD: 3.46-5.03), contemporaneous with the formation of Kauai. Interestingly, this age is also consistent with the approximate colonization date for plants in the silversword alliance at ~5 Ma (Baldwin and Sanderson 1998). Philodoria radiated onto members of the silversword alliance twice, once simultaneous with the

silversword alliance arrival in Hawaii, and once again with a moth clade that is currently restricted to Maui. We believe that Philodoria tracked these host plants to other islands as they formed and diversified in the process.

Although other Philodoria species-groups in this study do not have age estimates for their host plant, our data depict similar patterns of colonization from older to newer islands. Urticaceae-feeding Philodoria (6 Ma (95% HPD: 4.65-6.58) evolved after the formation of Kauai, likely tracking their host plants as new islands formed over the hotspot. Radiation onto members of the plant family Myrtaceae occurred at 2 Ma (95%

HPD: 1.75-2.83), which is nearly consistent with age estimates of Metrosideros (0.5-1

Ma, (Wright et al. 2001), the dominant Hawaiian canopy plant genus and a colonist of young lava flows. Morden and Harbin (2013) found that Hesperomannia likely colonized

Hawaii as early as 2.3 Ma. Philodoria shifted onto this plant genus at 4 Ma (95% HPD:

2.91-3.93). The divergence between the two Philodoria that feed on Hesperomannia in the present study is compatible with their dates of colonization and hypothesized dispersal of the plant genus amongst the Islands (Morden and Harbin 2013).

The phylogenetic position of two host shift events onto plants in Malvaceae (P. hauicola and P. marginestrigata, Figure 4-5) suggests that colonization of this plant family was a pathway to additional host switching. The present-day ecology of both host plants, Hibiscus tilaceus and Sida fallax, supports this hypothesis. Each is widespread in lowland areas of the Islands today. Host plants associated with the Philodoria species that are immediately derived from each of these two basal host shifts are known to co- occur with H. tilaceus and S. fallax.

Conservation of Philodoria Moths and Their Host Plants

Because of their strict diet, the Philodoria species that rely exclusively on threatened or endangered Hawaiian plants should be the focus of future work on the genus. Our strongly-supported phylogenomic and associated host plant data provide a framework for more detailed conservation studies of these threatened Philodoria at the population level.

Of the 33 Philodoria taxa included in this study, eight feed exclusively on threatened or endangered Hawaiian plant species (IUCN 2013, U.S. Fish and Wildlife

Service 2015). Given that the field collections as part of the present study yielded seven undescribed Philodoria species, and that recent taxonomic work on the genus nearly doubled the described diversity of the genus (Kobayashi et al. in prep), future surveys are likely to find additional undescribed Philodoria species that have evaded sampling due to specialization on rare plants with narrow ranges. Philodoria host associations are conserved at the plant family level, which should also direct future field surveys for yet unknown Philodoria host plants.

APPENDIX A ACCESSION AND PARTITION TABLES

Table A-1. GenBank accession numbers for ingroup and outgroup taxa. Taxon CO1 EF-1α CAD CJ-049 KT982402 KT982415 KT982390 CJ-061 KT982403 KT982416 KT982391 CJ-064 KT982404 KT982417 KT982392 CJ-065 KT982405 KT982418 KT982393 CJ-068 KT982406 KT982419 KT982394 CJ-072 KT982407 KT982420 KT982395 CJ-077 KT982408 KT982421 KT982396 CJ-101 KT982409 KT982422 KT982397 CJ-105 KT982410 KT982423 KT982398 CJ-112 KT982411 KT982424 KT982399 CJ-135 KT982412 KT982425 KT982400 CJ-142 KT982413 KT982426 N/A CJ-144 KT982414 KT982427 KT982401 Conopomorpha sp. N/A JN125115 JN125058 Epicephala relictella JF797232 JN125122 JN125066 Parectopa robiniella KF367667 JN125129 JN125083

Table A-2. Data set partitions and the corresponding best-fitting model of sequence evolution. Partition Partition contents Substitution model 1 CO1 1st position TrNef 2 CO1 2nd position F81 3 CO1 3rd position HKY + G 4 EF-1α 1st position TVM + I 5 EF-1α 2nd position F81 6 EF-1α 3rd position HKY + I 7 CAD 1st position HKY + I 8 CAD 2nd position GTR + I 9 CAD 3rd position HKY + I

APPENDIX B PARTITION TABLES

Table B-1. Dataset 1 partition models. Asterisk next to partition number indicates most likely model. Partition Model LogL AIC w-AIC AICc w-AICc BIC w-BIC 1 GTR+R4 -30436.6487 61055.2974 + 0.8509 61059.4185 + 0.8458 61631.5156 + 0.9987 2 GTR+R4 -128321.6746 256825.3493 + 0.7636 256826.1379 + 0.7402 257550.3998 + 0.9999 3 GTR+R3 -36756.4224 73690.8448 + 0.8215 73693.8651 + 0.8116 74277.6255 + 0.9997 4 TN+R3 -9477.0618 19126.1237 + 0.369 19137.6433 + 0.4096 19576.2629 + 0.8722 5 GTR+I+G4 -26137.2456 52448.4912 + 0.4937 52452.5111 + 0.5157 52993.8039 + 0.9984 6 GTR+R4 -44560.0953 89302.1905 + 0.8908 89304.5245 + 0.8822 89929.268 + 0.9983 7 GTR+R3 -37304.89 74787.7801 + 0.8469 74790.7451 + 0.838 75376.1795 + 0.9994 8 GTR+R3 -60589.8192 121357.6385 - 0.0048 121359.4032 - 0.0043 121991.6277 + 0.8548 9 TIM2+R3 -43765.3118 87704.6235 - 0.0109 87706.6952 - 0.0112 88306.6509 + 0.9275 10 GTR+R3 -48180.7768 96539.5536 + 0.8737 96541.7684 + 0.8637 97153.5446 + 0.9998 11 TN+I+G4 -19719.3488 39606.6976 - 0.0001 39610.6687 - 0.0001 40128.4165 + 0.7021 12 GTR+R4 -37432.3741 75046.7481 + 0.8701 75049.2721 + 0.8608 75666.7981 + 0.6582 13 TN+I+G4 -11875.6857 23919.3714 + 0.0157 23926.1359 + 0.0195 24397.7003 + 0.8467 14 TIM2+R4 -18187.8149 36553.6298 - 0.0013 36560.2716 - 0.0016 37072.041 + 0.8072 15 TIM2+R4 -34903.2982 69984.5965 + 0.2876 69987.8989 + 0.2998 70563.5702 + 0.9967 16 TIM2+R3 -28034.397 56242.7941 - 0.006 56246.624 - 0.0059 56792.2271 + 0.8182 17 GTR+R3 -15337.4393 30852.8785 + 0.8802 30857.8599 + 0.8787 31396.0895 + 0.8563 18 TIM2+G4 -22237.685 44643.3701 - 0.0001 44646.9743 - 0.0001 45173.0515 + 0.9309 19 TIM2+R3 -28391.1521 56956.3042 - 0.0021 56960.2803 - 0.0023 57502.5495 + 0.5571 20 TIM+I+G4 -7150.3698 14470.7396 - 0.0004 14484.4932 - 0.001 14899.7047 + 0.3124 21 HKY+R3 -1545.7818 3221.5636 + 0.0215 3277.6421 + 0.1196 3441.8533 + 0.2091 22 TIM3+I+G4 -18863.7924 37897.5847 - 0.0001 37902.5558 - 0.0002 38408.8875 + 0.7111 23 TPM2u+R3 -8947.4395 18066.8789 + 0.0276 18078.9272 + 0.0386 18513.4066 + 0.2963 24 GTR+I+G4 -15443.9859 31061.9719 + 0.5686 31070.1426 + 0.6166 31547.5815 + 0.5799 25 K3Pu+R4 -12599.1115 25374.2231 + 0.032 25382.64 + 0.0412 25864.8745 + 0.6353 26 TN+R3 -3998.911 8169.8221 + 0.3357 8187.1415 + 0.4038 8587.5683 + 0.8686 27 TPM2u+I+G4 -2115.2733 4382.5467 + 0.1238 4451.8011 + 0.3258 4648.9519 + 0.357 28* TPM2u+I+G4 -663.8616 1479.7231 + 0.0411 1593.3542 + 0.1639 1722.3879 + 0.2019

Table B-2. Dataset 2 partition models. Asterisk next to partition number indicates most likely model. Partition Model LogL AIC w-AIC AICc w-AICc BIC w-BIC 1 GTR+R4 -35213.1965 70608.3929 + 0.8504 70611.6864 + 0.8425 71204.6056 + 0.8519 2 GTR+R3 -38326.3322 76830.6644 - 0.0391 76832.921 + 0.036 77443.0117 + 0.969 3 TIM3+R4 -17254.077 34686.1541 + 0.0488 34692.5904 + 0.0563 35207.2631 + 0.9348 4 K3Pu+I+G4 -26323.8334 52815.6669 - 0.0035 52817.281 - 0.0037 53411.8481 + 0.6949 5 GTR+R3 -45854.0572 91886.1144 + 0.798 91888.294 + 0.7834 92501.5162 + 0.9997 6 GTR+R4 -58320.4477 116822.8953 + 0.1671 116824.862 + 0.1543 117465.3749 + 0.7365 7 TPM3u+I+G4 -22020.3976 44208.7951 - 0.0001 44211.1605 - 0.0001 44773.2452 + 0.3049 8 GTR+R4 -36501.4521 73184.9043 + 0.8641 73187.2523 + 0.8537 73811.4416 + 0.9974 9 HKY+I+G4 -7129.5706 14425.1412 - 0.0001 14436.2786 - 0.0002 14856.5284 + 0.3101 10 GTR+R4 -82401.8449 164985.6898 + 0.5782 164987.0297 + 0.5509 165662.7752 + 0.9996 11 TN+G4 -14494.13 29154.26 - 0.0004 29158.4436 - 0.0004 29663.591 + 0.4485 12 TIM3+R3 -37303.8303 74781.6606 + 0.6994 74784.109 + 0.701 75369.3412 + 0.9845 13 TPM2u+R3 -36314.9116 72801.8231 - 0.0129 72804.3811 + 0.0135 73377.0736 + 0.4731 14 GTR+R3 -53092.5965 106363.1931 + 0.8676 106364.9265 + 0.8557 106998.7603 + 0.9996 15 K3Pu+I+G4 -20632.7338 41433.4676 + 0.0755 41435.8124 + 0.079 41998.6418 + 0.6115 16 GTR+R4 -85519.0222 171220.0445 + 0.88 171221.2023 + 0.8673 171910.336 + 0.9999 17 GTR+R4 -69117.9946 138417.9892 + 0.9091 138419.4335 + 0.8998 139088.3008 + 0.9923 18 K3Pu+R3 -19999.192 40170.3841 - 0.004 40173.3234 - 0.0043 40733.8711 + 0.8094 19 TIM3+R3 -15590.4078 31354.8155 + 0.7819 31359.3363 + 0.7869 31890.157 + 0.8229 20 TIM2e+R3 -35461.674 71091.348 - 0 71093.5252 - 0 71662.6714 + 0.9259 21* K3Pu+I+G4 -2501.8789 5155.7577 + 0.0738 5223.8043 + 0.403 5423.0842 + 0.503 22 TVM+R3 -37837.3629 75850.7257 - 0.0003 75852.4391 - 0.0002 76478.1941 + 0.6514 23 GTR+R3 -31209.483 62596.9659 + 0.882 62600.0361 + 0.8751 63182.3163 + 0.9989 24 TPM3u+R3 -7867.75 15907.4999 + 0.0302 15917.2294 + 0.04 16371.3205 + 0.4036 25 TPM2+I+G4 -21626.0858 43420.1715 - 0.0001 43423.3107 - 0.0001 43961.2303 + 0.2098 26 HKY+R3 -8055.3411 16276.6823 + 0.1194 16289.7507 + 0.1468 16695.6984 + 0.7152 27 TIM+R3 -12801.6097 25777.2194 + 0.6481 25786.6713 + 0.6811 26250.7672 + 0.6034 28 K3Pu+R3 -42014.7578 84201.5156 - 0 84202.5716 - 0 84852.1119 + 0.6661 29 K3Pu+G4 -4535.5306 9217.0612 + 0.0733 9223.7303 + 0.0909 9613.8051 + 0.7383 30 K3P+G4 -4633.9496 9423.8992 - 0.0001 9431.8759 - 0.0002 9844.5257 + 0.8133 31 K3Pu+I+G4 -11284.308 22728.6161 - 0.0005 22732.7474 - 0.0005 23214.7373 + 0.2359 32 TIM2+I+G4 -11088.052 22342.1039 - 0.0005 22346.9305 - 0.0006 22839.8821 + 0.8297 33 HKY+G4 -4650.0448 9444.0895 - 0 9452.2447 - 0 9819.6926 + 0.2287 34 TIM2+R3 -13501.0666 27176.1333 + 0.626 27179.6228 + 0.6367 27733.4975 + 0.3434 35 K3Pu+I+G4 -4019.0488 8194.0975 + 0.0288 8201.364 + 0.0398 8621.6589 + 0.5785 36 TIM3e+I+G4 -3806.8297 7765.6594 - 0.001 7779.6762 - 0.0025 8131.6481 + 0.2974 37 K3P+G4 -3415.9166 6975.8332 - 0 6984.7492 - 0 7345.373 + 0.6005 38 HKY+I+G4 -3280.1809 6706.3618 - 0.0031 6724.3385 + 0.0087 7035.9358 + 0.3506 39 GTR+R4 -381649.1634 763480.3267 + 0.8258 763480.6019 + 0.805 764300.9324 + 0.7592

Figure B-1. ASTRAL tree. Red numbers show ASTRAL bootstrap support. Blue boxes indicate full support.

Figure B-2. ML tree topology from 50-locus subsampled Robinson-Foulds dataset.

LIST OF REFERENCES

Asquith A. 1995. Alien species and the extinction crisis of Hawaii’s invertebrates. Endanger. Species Updat. 12:1–8.

Athens SJ, Ward J V., Blimm DW. 2007. Vegetation History of Laysan Island, Northwestern Hawaiian Islands. Pacific Sci. 61:17–37.

Baldwin BG, Sanderson MJ. 1998. Age and rate of diversification of the Hawaiian silversword alliance (Compositae). Proc. Natl. Acad. Sci. U. S. A. 95:9402–9406.

Bess EC, Catanach TA, Johnson KP. 2014. The importance of molecular dating analyses for inferring Hawaiian biogeographical history: a case study with bark lice (Psocidae: Ptycta). J. Biogeogr. 41:158–167.

Biomatters. 2013. Geneious Pro.

Breinholt JW, Earl C, Lemmon AR, Moriarty Lemmon E, Xiao L, Kawahara AY. 2017. Resolving relationships among the megadiverse butterflies and moths with a novel pipeline for Anchored Phylogenomics. Syst. Biol. 0:1–16.

Carr GD. 1985. Monography of the Hawaiian Madiinae (Asteraceae): Argyroxiphium, Dubautia and Wilkesia. Allertonia 4:1–123.

Clague DA. 1996. Growth and subsidence of the Hawaiian-Emperor volcanic chain. In: The Origin and Evolution of Pacific island Biotas, New Guinea to Eastern Polynesia: Patterns and Processes. Amsterdam: SPB Academic Press. p. 35–50.

Clague D, Braga JC, Bassi D, Fullagar PD, Renema W, Webster JM. 2010. The maximum age of Hawaiian terrestrial lineages: geological constraints from Kōko Seamount. J. Biogeogr.

Conant S, Christensen CC, Conant P, Gagne WC, Goff ML. 1984. The unique terrestrial biota of the Northwestern Hawaiian Islands. Proc. Res. Inv. NWHI.

Crane PRR, Jarzembowski E a. A. 1980. Insect leaf mines from the Palaeocene of southern England. J. Nat. Hist. 14:629–636.

Department of Land and Natural Resources. State of Hawai‘i. 2007. Ka‘ena Point Ecosystem Restoration Project: Draft Environmental Assessment.

Drummond AJ, Suchard MA, Xie D, Rambaut A. 2012. Bayesian phylogenetics with BEAUti and the BEAST 1.7. Mol. Biol. Evol. 29:1969–1973.

Edgar RC. 2004. MUSCLE: multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 32:1792–7.

Eldredge LG, Evenhuis NL. 2002. Records of the Hawaii Biological Survey for 2000. Bish. Museum Occas. Pap. 68:71–78.

Feinberg M, Willer R. 2013. The Moral Roots of Environmental Attitudes. Psychol. Sci. 24:56–62.

Filipski A, Murillo O, Freydenzon A, Tamura K, Kumar S. 2014. Prospects for building large timetrees using molecular data with incomplete gene coverage among species. Mol. Biol. Evol. 31:2542–2550.

Gagné W. 1988. Conservation priorities in Hawaiian natural systems. Bioscience 38:264–271.

Givnish TJ, Millam KC, Mast AR, Paterson TB, Theim TJ, Hipp AL, Henss JM, Smith JF, Wood KR, Sytsma KJ. 2009. Origin, adaptive radiation and diversification of the Hawaiian lobeliads (Asterales: Campanulaceae). Proc. Biol. Sci. 276:407–16.

Goodman KR, Evenhuis NL, Bartosova-Sojkova P, O’Grady PM. 2014. Diversification in Hawaiian long-legged flies (Diptera: Dolichopodidae: Campsicnemus): Biogeographic isolation and ecological adaptation. Mol. Phylogenet. Evol. 81:232–241.

Guindon S, Dufayard JF, Lefort V, Anisimova M, Hordijk W, Gascuel O. 2010. New algorithms and methods to estimate maximum-likelihood phylogenies: Assessing the performance of PhyML 3.0. Syst. Biol. 59:307–321.

Haines WP, Schmitz P, Rubinoff D. 2014. Ancient diversification of Hyposmocoma moths in Hawaii. Nat Commun 5:3502.

Hao G, Yuan YM, Hu CM, Ge XJ, Zhao NX. 2004. Molecular phylogeny of Lysimachia (Myrsinaceae) based on chloroplast trnL-F and nuclear ribosomal ITS sequences. Mol. Phylogenet. Evol. 31:323–339.

Heads M. 2011. Old taxa on young Islands: A critique of the use of Island age to date Island-endemic clades and calibrate phylogenies. Syst. Biol. 60:204–218.

Hebert P, Penton E, Burns J, Janzen D, Et Al. 2004. Ten species in one: DNA barcoding reveals cryptic species in the neotropical skipper butterfly. Proc. Natl. Acad. Sci.

Hennig W. 1966. Phylogenetic Systematics. Annu. Rev. Entomol. 10:97–116.

Ho SYW, Tong KJ, Foster CSP, Ritchie AM, Lo N, Crisp MD. 2015. Biogeographic calibrations for the molecular clock. Biol. Lett. 11:20150194.

Howarth FG, Gangé B. 2012. Development of Insect Conservation in Hawai‘i. New TR, editor. Springer.

Howarth FG, Mull W. 1992. Hawaiian insects and their kin. Honolulu, HI: University of Hawaii Press.

Illumina. 2014. TruSeq RNA Low Sample Protocol v2.

IUCN. 2013. IUCN Red List of Threatened Species. Version 2013.1. IUCN 2013.

Johns CA, Moore MR, Kawahara AY. 2016. Molecular Phylogeny, Revised Higher Classification, and Implications for Conservation of Endangered Hawaiian Leaf-Mining Moths (Lepidoptera: Gracillariidae: Philodoria). Pacific Sci. 70:361–372.

Johns CA, Tangalin N, Bustamente K, Kawahara AY. 2014. Evidence of an Undescribed, Extinct Philodoria Species (Lepidoptera: Gracillariidae) from Hawaiian Hesperomannia Herbarium Specimens. Proc. Hawaiian Entomol. Soc. 46:55–57.

Jordan S, Simon C, Polhemus D. 2003. Molecular Systematics and Adaptive Radiation of Hawaii’s Endemic Damselfly Genus Megalagrion (Odonata: Coenagrionidae). Syst. Biol. 52:89–109.

Kahn PH, Kellert SR. 2002. Children and nature: psychological, sociocultural, and evolutionary investigations. Cambridge, MA: MIT Press.

Kahn PH, Severson RL, Ruckert JH. 2009. The human relation with nature and technological nature. Curr. Dir. Psychol. Sci. 18:37–42.

Katoh K, Standley DM. 2013. MAFFT multiple sequence alignment software version 7: Improvements in performance and usability. Mol. Biol. Evol. 30:772–780.

Katoh T, Izumitani HF, Yamashita S, Watada M. 2017. Multiple origins of Hawaiian drosophilids: Phylogeography of Scaptomyza Hardy (Diptera: Drosophilidae). Entomol. Sci.:33–44.

Kawahara AY, Breinholt JW, Ponce F V., Haxaire J, Xiao L, Lamarre GPA, Rubinoff D, Kitching IJ. 2013. Evolution of Manduca sexta hornworms and relatives: Biogeographical analysis reveals an ancestral diversification in Central America. Mol. Phylogenet. Evol. 68:381–386.

Kawahara AY, Ohshima I, Kawakita A, Regier JC, Mitter C, Cummings MP, Davis DR, Wagner DL, De Prins J, Lopez-Vaamonde C. 2011. Increased gene sampling strengthens support for higher-level groups within leaf-mining moths and relatives (Lepidoptera: Gracillariidae). BMC Evol. Biol. 11.

Kawahara AY, Plotkin D, Ohshima I, Lopez-Vaamonde C, Houlihan PR, Breinholt JW, Kawakita A, Xiao L, Regier JC, Davis DR, et al. 2017. A molecular phylogeny and revised higher-level classification for the leaf-mining moth family Gracillariidae and its implications for larval host-use evolution. Syst. Entomol.:60–81.

Kawahara AY, Rubinoff D. 2013. Convergent evolution of morphology and habitat use in the explosive Hawaiian fancy case caterpillar radiation. J. Evol. Biol. 26:1763–73.

Kawakita A, Kato M. 2009. Repeated independent evolution of obligate pollination mutualism in the Phyllantheae-Epicephala association. Proc. Biol. Sci. 276:417–26.

Kidwell SM, Holland SM. 2002. The quality of the fossil record: implications for evolutionary analysis. Annu. Rev. Ecol. Syst. 33:561–588.

Kono Y, Chung KF, Chen CH, Hoshi Y, Setoguchi H, Chou CH, Oginuma K, Peng CI. 2012. Intraspecific karyotypic polymorphism is highly concordant with allozyme variation in Lysimachia mauritiana (Primulaceae: Myrsinoideae) in Taiwan: Implications for the colonization history and dispersal patterns of coastal plants. Ann. Bot. 110:1119–1135.

Kraus F, Preston D. 2012. Diet of the Invasive Lizard Chamaeleo jacksonii (Squamata: Chamaeleonidae) at a Wet-Forest Site in Hawai’i. Pacific Sci. 66:397–404.

Lanfear R, Calcott B, Ho SYW, Guindon S. 2012. Partitionfinder: combined selection of partitioning schemes and substitution models for phylogenetic analyses. Mol. Biol. Evol. 29:1695–701.

Lanfear R, Calcott B, Kainer D, Mayer C, Stamatakis A. 2014. Selecting optimal partitioning schemes for phylogenomic datasets. BMC Evol. Biol. 14:82.

Lanfear R, Frandsen PB, Wright AM, Senfeld T, Calcott B. 2017. Partitionfinder 2: New methods for selecting partitioned models of evolution for molecular and morphological phylogenetic analyses. Mol. Biol. Evol. 34:772–773.

Lapoint RT, O’Grady PM, Whiteman NK. 2013. Diversification and dispersal of the Hawaiian Drosophilidae: The evolution of Scaptomyza. Mol. Phylogenet. Evol. 69:95– 108.

Lees DC, Lack HW, Rougerie R, Hernandez-Lopez A, Raus T, Avtzis ND, Augustin S, Lopez-Vaamonde C. 2011. Tracking origins of invasive herbivores through herbaria and archival DNA: the case of the horse-chestnut leaf miner. Front. Ecol. Environ. 9:322– 328.

Lemmon AR, Emme SA, Lemmon EM. 2012. Anchored hybrid enrichment for massively high-throughput phylogenomics. Syst. Biol. 61:727–744.

López-Vaamonde C, Godfray HCJ, Cook JM. 2003. Evolutionary Dynamics of Host- Plant Use in a Genus of Leaf-Mining Moths. Evolution (N. Y). 57:1804–1821.

Lozano-Fernandez J, dos Reis M, Donoghue PCJ, Pisani D. 2017. RelTime rates collapses to a strict clock when estimating the timeline of animal diversification. Genome Biol. Evol.

Maddison WP, Maddison DR. 2017. Mesquite: a modular system for evolutionary analysis.

Matzke NJ. 2013. BioGeoBEARS: Biogeography with Bayesian (and likelihood) Evolutionary Analysis in R scripts. R package, version 0.2, 1.

Medeiros MJ, Eiben JA, Haines WP, Kaholoaa RL, King CBA, Krushelnycky PD, Magnacca KN, Rubinoff D, Starr F, Starr K. 2013. The Importance of Insect Monitoring to Conservation Actions in Hawaii. Proc. Hawaiian Entomol. Soc.

Mello B, Schrago CG. 2012. Incorrect handling of calibration information in divergence time inference: An example from Volcanic Islands. Ecol. Evol. 2:493–500.

Meyrick E. 1912. Adelidae, Micropterygidae, Gracilariadae. In: Wagner H, editor. Lepidopterorum Catalogus 6. p. 25–68.

Meyrick E. 1928. Exotic Microlepidoptera.

Miller JR. 2005. Biodiversity conservation and the extinction of experience. Trends Ecol. Evol. 20:430–434.

Minh BQ, Nguyen MAT, Von Haeseler A. 2013. Ultrafast approximation for phylogenetic bootstrap. Mol. Biol. Evol. 30:1188–1195.

Mirarab S, Reaz R, Bayzid MS, Zimmermann T, S. Swenson M, Warnow T. 2014. ASTRAL: Genome-scale coalescent-based species tree estimation. Bioinformatics 30:541–548.

Misof B, Liu S, Meusemann K, Peters RS, Donath a., Mayer C, Frandsen PB, Ware J, Flouri T, Beutel RG, et al. 2014. Phylogenomics resolves the timing and pattern of insect evolution. Science 346:763–767.

Monteiro A, Pierce NE. 2001. Phylogeny of Bicyclus (Lepidoptera: Nymphalidae) Inferred from COI, COII, and EF-1α Gene Sequences. Mol. Phylogenet. Evol.

Morden CW, Harbin SC. 2013. Evolution and Biogeographic Origins of the Endemic Hawaiian Genusn Hesperomannia (Asteraceae). Pacific Sci. 67:219–235.

Nadkarni NM, Hasbach PH, Thys T, Crockett EG, Schnacker L. 2017. Impacts of nature imagery on people in severely nature-deprived environments. Front. Ecol. Environ. 15:395–403.

Nguyen LT, Schmidt HA, Von Haeseler A, Minh BQ. 2015. IQ-TREE: A fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies. Mol. Biol. Evol. 32:268–274.

Opler PA. 1974. Oaks as Evolutionary Islands for Leaf-Mining Insects. Am. Sci. 62:67– 73.

Oppenheimer HL, Bustamente KM, Perlman SP. 2014. A new species of Hibiscadelphus Rock (Malvaceae, Hibisceae) from Maui, Hawaiian Islands. PhytoKeys 75:65–75.

Price JP, Clague DA. 2002. How old is the Hawaiian biota? Geology and phylogeny suggest recent divergence. Proc. R. Soc. B Biol. Sci. 269:2429–35.

Pyle RM. 1978. The extinction of experience. Horticulture 56:64–67.

Rambaut A, Drummond A. 2009. FigTree.

Rauzon MJ. 2001. Isles of refuge: Wildlife and history of the Northwestern Hawaiian Islands. Honolulu, HI: University of Hawai‘i Press.

Ree RH, Moore BR, Webb CO, Donoghue MJ. 2005. A likelihood framework for inferring the evolution of geographic range on phylogenetic trees. Evolution 59:2299– 311.

Ree RH, Smith SA. 2008. Maximum likelihood inference of geographic range evolution by dispersal, local extinction, and cladogenesis. Syst. Biol. 57:4–14.

Robinson DF, Foulds LR. 1981. Comparison of phylogenetic trees. Math. Biosci. 53:131–147.

Robinson O, Dylus D, Dessimoz C. 2016. Phylo.io: Interactive Viewing and Comparison of Large Phylogenetic Trees on the Web. Mol. Biol. Evol. 33:2163–2166.

Ronquist F, Teslenko M, van der Mark P, Ayres DL, Darling A, Hohna S, Larget B, Liu L, Suchard M a., Huelsenbeck JP, et al. 2012. MrBayes 3.2: Efficient Bayesian Phylogenetic Inference and Model Choice Across a Large Model Space. Syst. Biol. 61:539–542.

Rubinoff D, Jose MS. 2010. Life history and host range of Hawaii’s endangered Blackburn’s sphinx moth (Manduca blackburni Butler). Proc. Hawiian Entomol. Soc. 42:53–59.

Russo CA, Takezaki N, Nei M. 1995. Molecular phylogeny and divergence times of drosophilid species. Mol. Biol. Evol. 12:391–404.

Shimodaira H, Hasegawa M. 2001. CONSEL: for assessing the confidence of phylogenetic tree selection. Bioinformatics.

Sohn JC, Regier JC, Mitter C, Davis D, Landry JF, Zwick A, Cummings MP. 2013. A Molecular Phylogeny for Yponomeutoidea (Insecta, Lepidoptera, Ditrysia) and Its Implications for Classification, Biogeography and the Evolution of Host Plant Use. PLoS One 8.

Stamatakis A. 2014. RAxML version 8: a tool for phylogenetic analysis and post- analysis of large phylogenies. Bioinformatics 30:1312–1313.

Sul SJ, Williams TL. 2008. An experimental analysis of robinson-foulds distance matrix algorithms. Lect. Notes Comput. Sci. (including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics) 5193 LNCS:793–804.

Swafford D. 2003. PAUP*.

Swezey OH. 1910. Some new species of Hawaiian Lepidoptera. Proc. Hawaiian Entomol. Soc. 2:103–106.

Swezey OH. 1915. New species of Hawaiian moths. Proc. Hawaiian Entomol. Soc. 3:93–97.

Swezey OH. 1940. New Species of Hawaiian Lepidoptera. Proc. Hawaiian Entomol. Soc.:523–525.

Swezey OH. 1954. Forest Entomology in Hawaii.

Toussaint EFA, Short AEZ. 2017. Biogeographic mirages? Molecular evidence for dispersal-driven evolution in Hydrobiusini water scavenger beetles. Syst. Entomol.

Townsend M, Weerasuriya R. 2010. Beyond Blue to Green: The benefits of contact with nature for mental health and well-being.

U.S. Fish and Wildlife Service. 2015. U.S. Fish and Wildlife Service Environmental Conservation Online System.

Wagner W, Bruegmann M, Herbst D, Lau J. 1999. Hawaiian vascular plants at risk. Bish. Museum Occas. Pap. 60:1–58.

Wagner W, Herbst D, Sohmer S. 1990. Manual of the Flowering Plants of Hawaii, Revised Edition. University of Hawaii Press.

Wahlberg N, Wheat CW, Peña C. 2013. Timing and patterns in the taxonomic diversification of Lepidoptera (butterflies and moths). PLoS One 8:1–8.

Walsingham, Lord. 1907. Fauna Hawaiiensis. Cambridge: The University Press.

Wright SD, Yong CG, Wichman S, Dawson JW, Gardner RC. 2001. Stepping stones to Hawaii: a trans-equatorial dispersal pathway for Metrosideros (Myrtaceae) inferred from nrDNA (ITS+ETS). J. Biogeogr. 28:769–774.

Xia X, Xie Z, Salemi M, Chen L, Wang Y. 2003. An index of substitution saturation and its application. Mol. Phylogenet. Evol. 26:1–7.

Zheng Y, Weins J. 2015. Do missing data influence the accuracy of divergence-time estimation with BEAST? Mol. Phylogenet. Evol. 85:41–49.

Zimmerman EC. 1970. Adaptive radiation in Hawaii with special reference to insects. Biotropica 2:32–38.

Zimmerman EC. 1978. Insects of Hawaii. Volume 9. Microlepidoptera, Part 1. Gracillarioidea. Honolulu: University of Hawaii Press.

BIOGRAPHICAL SKETCH

Chris A. Johns received his doctorate from the Department of Biology at the

University of Florida in fall 2017. He is from Gainesville, Florida and loves his home. He

also cares deeply about the Hawaiian Islands and culture. Chris is also Filipino-

American and takes pride in his heritage. He is inspired by design, nature, and creative

communication.