Genetic Structure and Colonization of North America by Depressaria depressana (Fabricius 1775) (Lepidoptera: Depressariidae) over 15 Years; Contrasts with Westward Expansion of Depressaria radiella (Goeze, 1783) over 160 Years

Simple Summary As invasive insect species can disrupt agriculture, ecosystem functions, and human health, monitoring the spread of newly invasive insects is a high priority. The purple carrot-seed moth Depressaria depressana, a European species first reported in North America in 2008, specializes on reproductive structures of taxa throughout the family Apiaceae and thus presents a potential threat to native North American umbellifers. We assessed interpopulational genetic diversity of D. depressana across its eastern North American invasion front and compared those values to available data for European populations. We also compared this diversity with genetic diversity estimates for populations of Depressaria radiella (parsnip webworm), a European species introduced to North America more than 160 years ago that is essentially restricted to two apiaceous genera throughout its native and invasive ranges. Documenting the historical spread of D. radiella with museum and literature records, we found D. depressana displays greater genetic diversity than D. radiella and is colonizing North America more rapidly; greater genetic diversity may facilitate faster colonization via establishment on a wider range of hostplants. The contrast in colonization by these two species may inform management practices for invasive insects utilizing a broad versus narrow array of hosts within a single plant family. Abstract Depressaria depressana, the purple carrot seed moth, is a Eurasian species first reported in North America in 2008 and currently undergoing range expansion. This invasion follows that of its Eurasion congener Depressaria radiella (parsnip webworm), first documented in North America 160 years ago. Unlike D. depressana, which utilizes hostplants across multiple tribes of Apiaceae, Depressaria radiella is a “superspecialist” effectively restricted in its native and non-indigenous ranges to two closely related apiaceous genera. We investigated the genetic structure of D. depressana populations across latitudinal and longitudinal gradients in the eastern United States by constructing COI haplotype networks and then comparing these with haplotype networks constructed from available COI sequence data from contemporary European D. depressana populations and from European and North American D. radiella populations. Haplotype data revealed higher genetic diversity in D. depressana, indicating high dispersal capacity, multiple introductions, and/or a genetically diverse founding population. Museum and literature records of D. radiella date back to 1862 and indicate that range expansion to the West Coast required more than 50 years. Higher levels of genetic diversity observed in D. depressana compared to its congener may indicate a greater propensity for dispersal, colonization and establishment in its non-indigenous range.


Introduction
A substantial increase in the number of invasive species in North America has occurred over the past 50 years as a direct result of increased globalization of trade and travel [1][2][3]. Managing the introduction of exotic insect species into new habitats is particularly problematic owing to their typical ease of transport, adaptability to new environments, rapid generation times, and high reproductive capacity [4,5]. In North America alone, there are more than 470 introduced insect species (Center for Invasive Species and Ecosystem Health, invasive.org), which annually incur costs exceeding $23 billion in the region [6]. This damage includes substantial loss of native biodiversity, loss of ecosystem services, crop yield loss, destruction of infrastructure, and declines in human health. Developing successful prevention and management practices is predicated upon a thorough understanding of both the ecological and evolutionary processes underlying insect colonization of novel habitats. To that end, monitoring the spread of a recently introduced species offers a unique opportunity to develop models of insect dispersal across, and adaptation to new environments.
The purple carrot-seed moth Depressaria depressana (Fabricius 1775) (Lepidoptera: Depressariidae), native to Eurasia, was first documented in the scientific literature as an introduced species in the United States in 2015 [7] and has likely been established in North America since at least 2008 [8], with images of larvae and adults appearing on BugGuide from New York, Connecticut, and Massachusetts as early as 2010. Although restricted to hostplants in the family Apiaceae, D. depressana utilizes species across a wide taxonomic cross-section of the family [7]. Moreover, in contrast with most Depressaria species, it is multivoltine and exhibits seasonal activity from late spring to early fall in central Illinois (C. Dean and M. Berenbaum, personal observations).
In expanding its range to include North America, D. depressana follows its Eurasian congener Depressaria radiella (Goeze, 1783) (formerly Depressaria pastinacella Duponchel, 1838), which was first reported in the Western Hemisphere in Ontario in 1862 (Bethune 1869, named D. ontariella and subsequently synonymized with the European D. heracliana (Linnaeus, 1758)). Like D. depressana, D. radiella is associated with hostplants in the family Apiaceae, but it is restricted to the two closely related genera, Heracleum Linnaeus, 1753 and Pastinaca (Linnaeus 1753), in North America and is univoltine throughout its range. Early reports of its occurrence in North America were associated with the damage it inflicted on flowers and fruits of Pastinaca sativa (Linnaeus 1753), the edible parsnip [9][10][11][12][13]. P. sativa is an apiaceous weedy biennial native to Eurasia thought to have been domesticated as a root vegetable centuries ago. Both wild and domesticated forms were introduced to North America in the seventeenth century [14] and the range of the interaction between parsnip webworms and the wild parsnip now extends across much of Canada and the United States [14].
Examination of P. sativa sampled from herbarium collections revealed that North American wild parsnip populations increased in toxic furanocoumarin content shortly after the accidental introduction of D. radiella to North America [15] and that, within its midwestern US range, parsnip webworms exhibit population-level detoxification profiles reflecting the furanocoumarin profiles in the immature fruits of their corresponding hostplant populations [16]. The reestablishment of the D. radiella and P. sativa system across novel landscapes has since been repeated in New Zealand. In 2004, D. radiella was first documented in Dunedin roughly 150 years after the introduction of P. sativa [17]. Studies conducted over a six-year period after the introduction of D. radiella to New Zealand found that P. sativa fitness was adversely affected following a reduction in seed production by up to 75% in six previously uninfested populations. As a compensatory mechanism, P. sativa populations in New Zealand, which were distinct from European and North American populations in phytochemical profile, exhibited increased growth rate accompanied by greater production of the floral volatile octyl butyrate [18].
For more than 40 years, the relationship between wild parsnip and the parsnip webworm D. radiella has served as a useful model for studying exotic plant-insect reassociations within and across novel landscapes [19]. While the colonization of D. radiella across multiple continents provides an exemplar of rapid ecological and evolutionary change following establishment of an invasive species, it is inherently constrained by the tightly coevolved nature of the D. radiella-P. sativa relationship. Since its introduction approximately 160 years ago, for example, D. radiella has acquired only a single native hostplant, Heracleum maximum, American cow-parsnip, the sole North American congener of many of its European hostplants [20]. D. depressana, however, utilizes a broader array of host plants in its native range species and has been associated with at least a dozen additional apiaceous genera in a diversity of tribes across Eurasia (Dean et al. submitted). Moreover, the extended period of seasonal activity of D. depressana, combined with its status as a family level specialist, suggests that, while the selection pressure exerted on its many hostplants may be less intense than that exerted by its congener, the broader host range and extended period of seasonal activity suggest that this species has a greater likelihood of encountering and acquiring native North American plants as novel hosts (Dean et al. submitted). Investigating the colonization of the family specialist D. depressana across North America can serve as a unique comparative measure against that of the "superspecialist" D. radiella (restricted to only two genera) and may inform predictions of the likely establishment and potential selective impact of invasive herbivores as a function of diet breadth.
A critical component of tracking ongoing biological invasions is the assessment of populational genetic structure along putative routes of establishment, which can facilitate inferences as to the source population, mode of introduction, and rate of evolution. In this study, we used the molecular marker mitochondrial cytochrome c oxidase subunit I (COI) to assess the population genetic structure of D. depressana in the eastern United States. Using this marker, we constructed haplotype networks to evaluate the inter-and intra-populational genetic diversity across both latitudinal and longitudinal gradients to the north and east of Illinois, where D. depressana was first reported in the United States [7]. We also constructed a haplotype network incorporating both European and North American D. radiella, using COI barcodes available in the BOLD [21] database. Lastly, using historical publications and museum records, we reconstructed an approximate timeline of the range expansion of D. radiella across North America. Together, the COI haplotype network and the approximate invasion timeline of the superspecialist D. radiella can provide a comparative insight as to whether the ecological differences in diet breadth and period of seasonal activity influence population structure across geographic space and whether these impacts are likely to influence the rate of successful D. depressana establishment across North America.

D. depressana Sampling
D. depressana larvae were sampled from the umbels of wild carrot Daucus carota at eight different collection sites along a latitudinal gradient from Urbana, IL to Eau Claire, WI in July 2020 and nine collection sites along a longitudinal axis from Urbana, IL to Newburg, PA in July 2021. Sample sites were selected by distance and collections occurred every 1 h of driving, or approximately every 70 miles (113 km). Ten ultimate instar larvae were collected from each site and preserved in labeled glass tubes filled with 90% EtOH. The state and GPS coordinates corresponding to each collection site are given in Table 1. These GPS locations are presented in maps in Figure 1.

DNA Extractions and PCR Amplification
For freshly collected specimens, total genomic DNA was collected from whole-body tissue extractions of D. depressana using the Qiagen DNeasy Blood and Tissue Kit (Qiagen Inc., Valencia, CA, USA) following the manufacturer's protocol. Ultimate instar caterpillars were individually homogenized in a solution containing buffer ATL and proteinase K using an autoclavable pestle (Fisher Scientific, Hampton, NH, USA). The homogenized tissue was incubated at 56 • C for 1 h to ensure complete cell lysis. A series of wash buffers was used to remove RNA, proteins, and other contaminants from the solution before DNA was ultimately eluted in 50 µL of ddH 2 O. DNA concentration was assessed via Nanodrop spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).
PCR amplification was performed for mitochondrial COI. The partial sequence of COI was amplified with the following primers from  Additionally, COI barcodes were sourced from the BOLD database [21] to be included in the COI haplotype network as a comparative measure to evaluate potential source populations. Similarly, COI barcodes for D. radiella were also obtained from BOLD for reconstruction of European and North American haplotypes (http://v4.boldsystems.org/ index.php/Public_BINSearch?searchtype=records, accessed on 25 June 2022). BOLD IDs and Genbank accession numbers for all COI barcodes generated and retrieved as part of this analysis are given in Supplementary Materials Table S1.

Alignment and Gene Characterization
Raw sequence reads were imported to Geneious v.2021.1.1 (Dotmatics, Boston, MA, USA) and ends were trimmed using an error probability limit of 0.05. Complementary paired sequences were assembled de novo, read quality was visually assessed by distinctness of fluorescence peaks. The consensus regions were then extracted, and primer regions were trimmed. All COI sequences were aligned in Geneious v.2021.1.1 (Dotmatics, Boston, MA, USA) via MAFFT v 7.450 [23] using default parameters.

Haplotype Network Construction and Analysis
Partial D. depressana and D. radiella sequences for COI were exported separately as Nexus files that included trait blocks incorporating sampling site assignments. Haplotype designations and networks were generated in PopART v.1.7 [24] using the Templeton-Crandall-Sing (TCS) algorithm [25]. Genetic structuring across sampling sites was assessed across geographic space via analysis of molecular variance (AMOVA) with PopART v.1.7 [24]. A Mantel test was also performed using zt v.1.1 [26] to evaluate the relationship between genetic variation and geographic distance.
Publications documenting the earliest observations of D. radiella in North American states and provinces were retrieved from PubMed.gov. Search terms "Depressaria pastinacella" "Depressaria heracliana" and "Depressaria ontariella" were used and the results were filtered by date. Additionally, physical copies of publications documenting early observations of D. radiella in North America were included in this analysis. These publications were retrieved from the private collection of one of the authors (MRB). Searches for D. radiella museum records were performed in iDigBio.org using the query terms "Depressaria radiella" or "Depressaria pastinacella" with United States and Canada as the "Country" specification. Search results were filtered by "Date" and "State/Province" of specimen collection. The earliest museum record of every state or province was used in the analysis and the corresponding museum was noted.

Depressaria Depressana Haplotypes
The COI locus for 167 D. depressana individuals was amplified and sequenced. Additionally, 34 sequences of COI were retrieved from the BOLD database and added to the alignment containing COI sequences generated in-house for this analysis. Across thẽ 680-km latitudinal gradient ranging from Urbana, IL, to Eau Claire, WI, 79 COI sequences were generated from D. depressana samples collected across eight geographic locations. COI sequences from the 34 additional taxa retrieved from the BOLD database were incorporated into the 658-bp nucleotide alignment. Uncorrected pairwise COI distances revealed genetic distances of 0-2%. Within this alignment, 22 polymorphic sites were identified as informative for the delineation of haplotypes, and individuals were subsequently clustered into eight distinct haplotypes ( Figure 2). A low level of genetic diversity was detected across all samples and the AMOVA revealed little genetic structuring (φ ST = 0.095, p = 0.059). clustered into eight distinct haplotypes (Figure 2). A low level of genetic diversity was detected across all samples and the AMOVA revealed little genetic structuring (ɸST = 0.095, p = 0.059). Analysis of the ~1078 km longitudinal gradient ranging from Urbana, IL, to Newburg, PA, incorporated 88 D. depressana samples collected from nine geographic locations. For these specimens, 88 COI sequences were generated and aligned to the 34 COI sequences retrieved from the BOLD database. Seven informative polymorphic sites were identified, and taxa were clustered into six distinct haplotypes. Similarly, a low level of genetic structure was observed (AMOVA, ɸST = 0.12, p = 0.068) (Figure 3). Analysis of the~1078 km longitudinal gradient ranging from Urbana, IL, to Newburg, PA, incorporated 88 D. depressana samples collected from nine geographic locations. For these specimens, 88 COI sequences were generated and aligned to the 34 COI sequences retrieved from the BOLD database. Seven informative polymorphic sites were identified, and taxa were clustered into six distinct haplotypes. Similarly, a low level of genetic structure was observed (AMOVA, φ ST = 0.12, p = 0.068) (Figure 3).  Combining the D. depressana COI datasets obtained from samples collected along the latitudinal and longitudinal gradients revealed 23 informative polymorphic sites, clustering the taxa into 5 haplotypes with 4 shared between gradients. An AMOVA run on the pooled dataset also revealed a low level of genetic structure (ɸST = 0.10, p = 0.058) ( Figure  4). The Mantel test returned a significant but very weak positive correlation between geographic distance and genetic variation (r = 0.049, p = 0.042), indicating that isolation by distance has little to no effect of genetic variation. Combining the D. depressana COI datasets obtained from samples collected along the latitudinal and longitudinal gradients revealed 23 informative polymorphic sites, clustering the taxa into 5 haplotypes with 4 shared between gradients. An AMOVA run on the pooled dataset also revealed a low level of genetic structure (φ ST = 0.10, p = 0.058) (Figure 4). The Mantel test returned a significant but very weak positive correlation between geographic distance and genetic variation (r = 0.049, p = 0.042), indicating that isolation by distance has little to no effect of genetic variation.

Depressaria radiella Haplotypes
With respect to variation at the COI barcode fragment in European and North American Depressaria radiella, analysis of the 658-bp region of COI for D. radiella incorporated 45 individuals. Twenty-six complete COI sequences were found, 16 from Europe and 10 from North America. In addition, 19 partial sequences were found, 6 from Europe and 13 from North America. The 26 complete DNA barcodes were aligned with NCBI Multiple Sequence Alignment ViewEr 1.22.0. It is clear from Table 2, the resulting haplotype network revealed two distinct haplotypes, with 5 informative SNPs distinguishing them (Table 2, Figure 5). All individuals fell into the H1 haplotype, with the single D. radiella specimen collected in Russia constituting the H2 haplotype.

Reconstruction of the History of Invasion of North America by D. radiella
The literature search results and museum records used in establishing the timeline of the westward expansion of D. radiella across North America are presented in Table 3

Reconstruction of the History of Invasion of North America by D. radiella
The literature search results and museum records used in establishing the timeline of the westward expansion of D. radiella across North America are presented in Table 3.

Discussion
COI is a widely used marker for assessing haplotype diversity in invasive and migratory Lepidoptera [27][28][29][30][31]. This gene evolves at a relatively slow rate in the class Insecta [32]. COI sequencing offers a reliable method of evaluating competing hypotheses regarding invasive populations and their introduction pathways into new geographic areas. Sampling both field-collected and museum specimens of Helicoverpa armigera across Argentina, Balbi et al. (2020) [27] uncovered 5 COI haplotypes, along with 3 cytochrome b (Cytb) haplotypes, which provided preliminary evidence of multiple points of introduction into the region from neighboring Latin American countries. Similarly, Lee et al. (2020) [31] proposed that, among populations of Spodoptera frugiperda (J.E. Smith), a species recently introduced to South Korea, the two distinct COI haplotypes present were suggestive of Brazil populations as the source of this invasion. This study was the first to monitor the genetic distribution of S. frugiperda, which, given its high capacity for migration and dispersal, makes it a candidate to become a major agricultural pest in South Korea [26]. In our study, we found 8 COI haplotypes among the eight sites spanning the latitudinal gradient and six haplotypes among the nine sites spanning the longitudinal gradient with 4 shared between gradients. Across all COI sequences, we found that populations of D. depressana established within the United States maintain relatively high levels of genetic diversity compared to D. radiella. The lack of strong genetic structure observed among sampling sites could also be an indication that the entirety of the sampling area is effectively one population (i.e., D. depressana is experiencing high gene-flow or rapid expansion). Although a larger sampling area might have been needed to detect genetic structure, these data clearly indicate that D. depressana are excellent dispersers. Multiple recent introductions throughout the country combined with high dispersal capacity would effectively obscure any genetic structure among sampling sites. The Mantel test further supports this assertion by providing significant evidence that geographic distance is has no major effect on patterns genetic variation-an indication that their propensity to disperse long distances is not driving genetic isolation.
Our findings are consistent with either multiple introductions of D. depressana or a single introduction of a large population retaining a large amount of diversity of ancestrally polymorphic alleles. While the dataset sampled from the BOLD database was limited, 6 haplotypes were recovered from the 35 sequences originating in Europe and Canada. However, the populations sampled in the United States do not strictly coincide to specific European or Canadian populations, consistent with ongoing interpopulational outcrossing, high levels of gene flow, or a lack of sufficient evolutionary time for geographic sorting of COI haplotypes. Multiple introductions may indeed provide an adaptive advantage to D. depressana, as it reduces the likelihood of bottleneck events.
In contrast, the COI haplotype network for D. radiella reveals relatively little genetic diversity within North America and across Europe, with significant geographic structuring between Russian and European/North American haplotypes. All individuals included in this analysis comprise a single haplotype except for one haplotype collected in Russia. This Russian individual was placed into a separate haplotype on the basis of four informative SNPs.
This apparent lack of variation, if validated by more extensive sampling, has significant implications for understanding the biology of D. radiella. Lack of variation in just the introduced North American populations could be explained by a simple, rapid loss of variation in a colonization bottleneck, but the low level of variation in the European source population would require a more complex evolutionary explanation, conceivably related to the moth's limited host range or dispersal capacity; the only information about dispersal capacity is reported by Zangerl and Berenbaum (2003) [33], who determined experimentally that gene flow via adult migration extends at least 1.3 km. However, before investing effort in such an evolutionary explanation, it is worth considering the Russian population in more detail. Several features stand out. First, the reported site, at 2600 m on Mt. Terskol in the Russian Caucasus mountains, is approximately 2000 km east of the rest of the European samples. Moreover, the wingspan of the specimen is 32 mm, well outside the reported wingspan range of D. radiella of 19-27 mm. Given the 5 unique bases found in this atypical specimen, the possibility that it is a new species-not D. radiella-should be given serious consideration.
If the Russian sample is removed, then all complete D. radiella sequences are identical. While entering a set of identical sequences into, e.g., PopArt returns a φFST value of "nil", for discussion such a φFST can be considered as 0 between the source (Europe) and colonized (North American) populations. Given this situation, the partial sequences might provide insights. For one thing, two additional haplotypes are revealed. The Croatian specimen and one of the Ontario, Canada specimens each has a unique haplotype (although these two haplotypes share the base change at position 628). These haplotypes indicate that further sequencing of D. radiella would reveal additional variation, although the available data suggest that such an increase in variation would be modest. Finally, the fact that one of the additional haplotypes is from Europe and the other is from North American suggests that more sampling could continue to show roughly equal amounts of variation in source and colonist samples.
Nevertheless, the apparent low genetic diversity may indicate a slower rate of molecular evolution in D. radiella compared to D. depressana. Multiple generations per year and a lengthier period of seasonal activity in D. depressana may be predicated upon its expanded host range, which comprises apiaceous hosts that have sequential yet overlapping blooming periods from May through September. It is possible that the larger number of hosts of D. depressana may have selected for greater genetic variation relative to D. radiella; evidence for the "niche width-variation hypothesis" is mixed, with some authors reporting increased variation in more polyphagous species [34,35] and others reporting the exact opposite relationship [36,37].
More than a century after the parsnip webworm first appeared in North America, reconstructing its history of westward and southward expansion is challenging, but, because of its status as an economic pest of cultivated parsnip, its first appearance in an area previously free of the pest often merited collection and preservation in a museum or a report in a journal. A survey of the literature and museum records relating to the expanding geographic distribution of the parsnip webworm reveals that D. radiella successfully expanded its range from the East Coast to the West coast in at most 52 years, between 1862 to 1914. In 1925, D. radiella was reported to have been established Yuma County in Arizona, bordering Mexico, which represents considerable southward expansion of its range (Clarke, 1941) [38]. By 1932, Dustan (1932) [39] noted that this species was present in "all provinces" of Canada. Given that P. sativa was a frequently planted crop in the early part of the D. radiella colonization of North America [40], it remains to be seen whether D. depressana will be able to expand its range at the same pace as D. radiella. Currently, the range of D. depressana extends west to Iowa, 14 years after its initial discovery in 2008 in Ontario. If Iowa is in fact the western edge only 14 years after its introduction, D. depressana is already outpacing D. radiella in colonizing North America. Such rapid range expansion may not be entirely attributable to biological traits; many more opportunities for rapid human-assisted dispersal exist today, for example, than existed for D. radiella in the 19th century. Notwithstanding, by virtue of its increased period of seasonal activity and the ubiquity of available hosts, particularly D. carota, along North American highways, D. depressana may indeed require less time to colonize the continent.
This study is unique in that D. depressana is still apparently in the early stages of its invasion, whereas many similar studies assess population structure long after invasive species have established (but see Nayyar et al., 2021) [41]. Invasive insects with relatively broad host ranges often have profound impacts on ecosystem dynamics, not only through imposing additional selection pressure on their food plants, but also in terms of competitive displacement of native insects and introduction of new pathogens [42]. Whereas D. radiella, a superspecialist herbivore, reestablished its tightly coevolved relationship with P. sativa in North America [18,40], subsequently colonizing a single new native host Heracleum maximum (a congener of European host species) [20] as far as is known, D. depressana offers a novel comparator for assessing the ecological impact of less specialized Lepidoptera. One avenue of exploration would be to monitor the impact of D. depressana on wild carrot, D. carota, a non-native but long-established umbellifer with low levels of furanocoumarin production [43]. Intense florivory by D. depressana may select for greater production of furanocoumarins, as D. radiella did in P. sativa within 20 years of its introduction to North America [15], potentially altering the community structure of native insects consuming D. carota and may similarly affect native umbellifers, such as golden alexanders, Zizia aurea, the seeds of which are consumed by this species in the laboratory (Dean et al., submitted).

Conclusions
In conclusion, comparing the invasion history of two congeneric species, which share some but not all ecological traits, provides a unique opportunity to gain insights into the extent to which certain life history traits affect the likelihood of acquiring novel hostplants in the area of introduction. Key shared ecological and behavioral traits include areas of indigeneity (i.e., Eurasia), utilization of hostplants in the Apiaceae, specialized consumption of flowers and immature fruits, and use of silken webbing to aggregate umbels. In contrast, differences in ecology and behavior between the two congeners include voltinism, per-plant densities within umbels, and diet breadth. Identifying which of these differences, if any, contribute to the degree of genetic differentiation across landscapes may have predictive power in anticipating the rate at which new communities are invaded and the extent of the threat they present to the native flora.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/insects13090789/s1, Table S1: Depressaria Specimen IDs.  Data Availability Statement: DNA sequence data analyzed in this study are publicly available in BOLD database and in Genbank.