Diversification in the microlepidopteran genus Mompha (Lepidoptera: Gelechioidea: Momphidae) is explained more by tissue specificity than host plant family

Insect herbivores and their hostplants constitute much of Earth’s described diversity, but how these often-specialized associations evolve and generate biodiversity is still not fully understood. We combined detailed hostplant data and comparative phylogenetic analyses of the lepidopteran family Momphidae to explore how shifts in the use of hostplant resources, not just hostplant taxonomy, contribute to the diversification of phytophagous insect lineages. We generated two phylogenies primarily from momphid species in the nominate genus, Mompha Hübner. A six-gene phylogeny was constructed with exemplars from Onagraceae hosts in western and southwestern USA and a cytochrome c oxidase subunit 1 (COI) phylogeny utilized both our collected sequences and publicly available accessions from the Barcode of Life Data System. Coalescent-based analyses combined with morphological data revealed ca. 56 undescribed Mompha species-level taxa, many of which are hostplant specialists on southwestern USA Onagraceae. Our phylogenetic reconstructions identified four major momphid clades: 1) an Onagraceae flower- and fruit-feeding clade, 2) a Melastomataceae galling clade, 3) an Onagraceae and Rubiaceae leafmining clade, and 4) a heterogeneous clade associated with multiple hostplant families, plant tissues, and larval feeding modes. Ancestral trait reconstructions on the COI tree identified leafmining on Onagraceae as the ancestral state for Momphidae. Cophylogenetic analyses detected loose phylogenetic tracking of hostplant taxa. Our study finds that shifts along three hostplant resource axes (hostplant taxon, plant tissue type, and larval feeding mode) contributed to the evolutionary success and diversification of Mompha. More transitions between exploited host tissue types than between hostplant families indicated that exploited host tissue (without a change in host) played an unexpectedly large role in the diversification of these moths.


Introduction
characterized by their small size and narrow forewings with raised scale tufts ( Fig 1A). Recent 72 studies of momphid genitalia and cytochrome c oxidase subunit 1 (COI) sequences indicate that 73 the family is rich with cryptic species and undescribed taxa, warranting more intensive study of 74 feeding niches, host plant usage, and systematic relationships across this group [28][29][30][31].  In this study, we combine phylogenetic reconstruction and detailed hostplant-resource 94 data (host taxonomy, plant tissue type, and larval feeding mode) to examine how shifts in 95 hostplant resource axes contributed to the diversification of a species-rich insect group. We 96 address the following questions: (1) What are the evolutionary relationships among Mompha? (2) 97 What is the ancestral hostplant taxon, tissue resource, and feeding mode?  From the collected Mompha, we randomly selected one exemplar from each feeding 116 resource axis: hostplant tissue type (flowers, fruits, leaves, shoot tips, stems, and roots), Mompha 117 feeding mode (galling, boring, or mining), and hostplant family ( For the 178 previously unsequenced samples, DNA was extracted from either the anterior 130 third of a caterpillar or a single adult leg, with a modified Chelex 100 and Proteinase-K protocol 131 (S1 appendix) [40]. We amplified partial coding sequences from one mitochondrial locus (COI) 132 and five nuclear loci [glyceraldehyde-3-phosphate dehydrogenase (GADPH), elongation factor  Table). These loci have been used to reconstruct species-level 135 relationships in other lepidopteran genera [41][42][43][44]. PCR was performed in 10 μL volumes (S1 136 appendix). Thermocycler programs were optimized for each primer pair (S1 Appendix). PCR 137 product was verified with a 1% agarose gel stained with SYBR® Safe DNA gel stain (Life   7   138 Technologies, Grand Island, NY, USA) and stored at 4°C. For failed reactions, PCR was 139 reattempted with internal primers generated for DDC, CAD, and GADPH (S2 Table). Successful 140 amplicons were purified using Exonuclease I and Shrimp Alkaline Phosphatase (Affymetrix) (S1 141 Appendix). Purified PCR product was sequenced in the forward direction using a modified 10 μL 142 BigDye® Terminator v3.1 (ThermoFisher) cycle sequencing reaction (S1 Appendix) with 143 standard thermocycler conditions (S1 Appendix). Sequenced product was purified with an 144 EtOH/EDTA cleanup (S1 Appendix) and visualized on an ABI 3730 High-Throughput 145 Sequencer.

146
Two datasets were generated for phylogenetic reconstruction: (1)  GenBank (S1 Table). 164 Phylogenies were reconstructed with Maximum Likelihood (ML), Bayesian inference, 165 and Coalescent methods using the CIPRES research computing resource [49]. Best  species-level taxa (Fig 2). Seventeen species-level taxa were recognized as undescribed: of these, 231 ten were collected in Central America and seven from the southwestern USA.  The combined COI dataset contained a maximum of 558 bp for 842 Mompha individuals.

240
ML and Bayesian phylogenies had highly similar, but non-congruent topologies 241 (10.5061/dryad.3n1g4td). The BEAST phylogeny was selected to represent the COI dataset 242 because it is ultrametric and most similar to the reconstructed phylogeny of the six-gene dataset.

243
As there were incongruences between the two datasets, we referred to the greater clade support 244 in the six-gene dataset to resolve conflicts in topology. Molecular delimitation for the COI 245 dataset recovered a range of 79-127 Mompha species-level taxa (S5 Table): GMYC: 87 taxa (CI:   Table). The COI phylogeny 272 identifies ten shifts to new hostplant families: one to Lythraceae, one to Polygonaceae, one to Lythraceae, and Cistaceae). Galling and boring taxa have taxonomically restricted host 278 associations, often feeding on a single hostplant species (see feeding records: S1 Table). Flower-   Ancestral trait reconstruction identified leafmining as the ancestral utilized hostplant 288 resource axis (Fig 5). Additional analyses that separate feeding mode and plant tissue type find 289 that mining was the ancestral feeding mode and that leaves were the ancestral feeding tissue type 290 (S1 and S2 Figs). The phylogeny identified 11 independent shifts from mining to new feeding 291 modes. The most common shifts were from boring to unknown, mining to unknown, mining to 292 boring, and from mining to galling (S7 Table). The COI phylogeny shows that Mompha shifted 293 to new hostplant tissue in 17 instances. Shifts from leaf to unknown, from flower to unknown, 294 and from leaf to stem + leaf were most prevalent (S8 Table). The Onagraceae-boring clade and 295 the Melastomataceae-galling clade had the highest observed rates of switching between hostplant 296 tissue types (Fig 3).  Mompha able to consume multiple hostplant species, sections, or sometimes genera (see feeding 313 records, S1 Table).     herbivores, will lag behind the diversification of their hosts [86].

403
In addition to onagrad taxonomic richness, Mompha niche partitioning on onagrad 404 hostplants appears to have contributed to diversification across this genus. Our phylogenies 405 document many instances of niche partitioning in Mompha, which allows for two or more Nearctic lineages [90,91]. In Mompha, we found that several Holarctic species complexes (e.g. into North American and European taxa. As was recently done with noctuid moths [92], 436 collaborations between European and North American Mompha experts will help resolve 437 species-level taxonomy of Holarctic taxa in this phenotypically challenging genus.

438
There remains many undescribed Mompha species in the Nearctic, Palearctic, and