Despite its well-known association with both maritime and inland dunes, the host(s) of S. riparia have eluded lepidopterists for decades. Among the candidate host plants suggested on the basis of their prevalence in dunes was American beachgrass (Ammophila breviligulata Fern.), which would account for a high concentration of records along the Atlantic Coast as well as the shores of the Great Lakes. Beachgrass represents not only an important structural and stabilizing component of dunes but an important host for herbivorous insect biodiversity, including other noctuid dune associates such as Apamea lintneri (Grote, 1873) (Quinter 2009: 103). With over 190 species in North America, Sympistis represents the most speciose noctuid genus on the continent, but relatively few species have documented larval life histories or host associations. Those which are known involve broad-leaved plants including Scrophulariaceae, Ericaceae, Caprifoliaceae and most particularly Rosaceae. Given this background, an association with a grass seemed highly unlikely: graminivory in Lepidoptera tends to be either highly conserved or associated with polyphagous feeding habits not known from any species of Sympistis. Feeding on grasses requires a complex of behaviors and morphological features necessary to consume and digest siliceous plant material, a suite of attributes unlikely to evolve frivolously. To our knowledge, no isolated species of graminivores are known to be nested within groups of folivores of broad-leaved plants, and an isolated Ammophila-feeder within a lineage dominated by Rosaceae feeders would be unique for Noctuidae, if not Lepidoptera.

With this in mind, we began our survey of potential host plants with what we surmised to be more likely candidates, the shadbushes or serviceberries Amelanchier Medikus. Amelanchier was also among the hosts recorded by Crumb (1956: 121) for A. nefascia (as Abagrotis n. sp.) (Lafontaine 1998). Although we recognized that S. riparia is unlikely to be strictly monophagous, it seemed plausible that the local distribution of Amelanchier in a range of coastal habitats, and the wider occurrence of various Amelanchier species would account for both the regional distribution of S. riparia and the more diffuse local distribution of adult moths. Based on our understanding of other Sympistis, we expected riparia to overwinter in the egg stage, and given that adults had been recorded as early as mid-June, we expected that a narrow larval development period was accommodated by feeding, at least initially, on floral tissues available in early spring. In this region, Amelanchier is the first flowering rosaceous shrub of the season, referred to as “shadbush” for the coincidence of its flowering with the return of shad to their freshwater spawning grounds.

As a state-listed species, observations of S. riparia are tracked by the Massachusetts Natural Heritage Program, and upon examining these data we initiated our survey at a series of sites near Moshup Trail in Aquinnah, Massachusetts (Figs 1–2), which at the time supported the highest known concentration of recent observations (Fig. 3a, b). This geologically unique area comprises a structurally complex mosaic of morainal heathlands, shrublands, and perched fens enclosing migrating dunes interspersed with cranberry (Vaccinium macrocarpon Aiton), sundew (Drosera intermedia Hayne) and bushy bluestem (Andropogon glomeratus). This unusual admixture of plant communities supports a unique assemblage of Lepidoptera (Goldstein, in prep.), including a number of regionally localized or unique occurrences. Among those species that had been recorded in significant numbers was A. nefascia, but given its wider distribution both in Massachusetts and on Martha’s Vineyard (Fig. 4a, b), we did not initially suspect it shared a primary host plant with S. riparia.

A casual survey for Sympistis larvae on Amelanchier in April 2011 was timed to coincide with the plant’s flowering season, but was unsuccessful. Later the same season, however, light trapping efforts at a site approximately three kilometers distant yielded adult moths in high numbers which were unprecedented in our experience. Although radically different from our initial target site in terms of its immediate plant community composition, which includes a mixture of mesic forest and morainal scrub, the trap site overlooks the expansive and under-studied peninsular dune system, Lobsterville Dunes. This site supports a structurally and floristically diverse vegetative cover lichenized and dominated by beach plum Prunus maritima, scrub oak Quercus ilicifolia Wangenh., and beach heather, Hudsonia tomentosa Nutt. (Fig. 5a–c), and pocketed with numerous graminoid and shrub-dominated wetlands and interdunal swales (Fig. 6a, b). The observation of high numbers of S. riparia in view of large, highly visible stands of beach plum led to the sudden, strong, and overdue suspicion that one of the most obvious native rosaceous dune plants had been overlooked as a potential host. It also led to our initiating survey efforts targeting beach plum caterpillars beginning the following spring (2012), at which time regular light trapping was also begun within the Lobsterville dune system itself, in parallel with ongoing light trapping at Moshup Trail. This was intended to evaluate whether S. riparia did indeed occur in consistently higher numbers within the strict confines of the dune system than had been observed previously. Over 100 light traps were deployed in Aquinnah, including 25 at Lobsterville Dunes between 2012 and 2016. Many specimens were prepared for study, and others frozen in liquid nitrogen for future molecular work.

Pinned specimens were examined with an incandescent light source. Genitalic preparations varied, the more recent ones following parts of Clarke (1941) and Lafontaine (2004), but using chlorazol black and mounted in euparol, the vesicae everted in water and fixed in ethanol. The most recent dissections were made following an overnight room-temperature soak in supersaturated sodium hydroxide, and examined prior to mounting. Slide preparations were examined with dissecting and compound microscopes. Photographs were made using the Microptics and Visionary Digital imaging systems and images manipulated with Adobe Photoshop® (Adobe Systems, Mountain View, CA). Measurements were made with the aid of an ocular micrometer. Forewing length was measured from the center of the axillary area up to the apex of the forewing (FW). Terminology follows Lafontaine (1998, 2004).

Even a cursory examination of the known distribution of Abagrotis nefascia reveals an obvious disjunction between the range occupied by nefascia west of the Rocky Mountains and the populations restricted to a narrow band along the northeastern coast of North America. Recognizing this, Franclemont (1955) attributed the north eastern populations as an infraspecific race (benjamini) of Abagrotis crumbi, which he described based on material reared by S.E. Crumb. Buckett retained both names and their ranks in his (1969c) revision of Abagrotis [part], but designated a specimen of A. crumbi as the lectotype of A. nefascia (Figs 13–18). Lafontaine synonymized both A. crumbi and A. crumbi benjamini with A. nefascia, having traced considerable confusion surrounding the identity of Abagrotis nefascia involving no fewer than four specific epithets: A. nefascia (Smith, 1908), A. forbesi (Benjamin, 1921), A. crumbi Franclemont, 1955, and A. reedi Buckett, 1969. As Lafontaine (1998: 221) recounts, A. nefascia had been predominantly associated with specimens referred to A. reedi by Buckett (1968b), who synonymized A. forbesi with A. nefascia. Although Buckett (1968b: 19) designated a lectotype of A. nefascia, the specimen is conspecific with Abagrotis crumbi and not with A. forbesi as he had supposed. Further confusing matters, the lectotype genitalia that was figured by Buckett (1969: fig. 88) included the aedeagus of the A. nefascia lectotype, but the mislabeled valvae belonging to a specimen of A. forbesi. In disentangling these threads, Lafontaine reversed Buckett’s synonymy, revived A. forbesi as a valid species and synonymized both A. crumbi and A. crumbi race benjamini with A. nefascia. The history of nomenclatural confusion, including the misspelling of A. nefascia as A. negascia in the original description, is summarized by Lafontaine (1998: 221): “Nefascia was treated as a senior synonym of forbesi by Buckett (1968c [cited herein as 1968b]) but the lectotype designated by Buckett (1968c[b]: 19) is clearly referable to the species previously known as crumbi. The valve shape and presence of a secondary diverticulum in the subbasal diverticulum of the lectotype are diagnostic for nefascia (=crumbi).”

Those populations formerly known as A. crumbi race benjamini have thus not been validly recognized as taxonomically distinct since 1998. Despite the formal synonymy, benjamini has been retained as a subspecies of A. nefascia outside the taxonomic literature, most prominently in the roster of protected species and subspecies in New York (New York Natural Heritage Program 2016), and Connecticut (Connecticut Dept. of Energy & Environmental Protection 2016) as Abagrotis nefascia benjamini, as well as at iNaturalist.org. This usage appears traceable to entries in NatureServe (http://explorer.natureserve.org/), a portal for disseminating information on taxa of conservation concern. Below we revisit the grounds for retaining or elevating the status of A. benjamini based on available character information, and revise its status accordingly, providing a diagnosis and re-description of the genitalia. A more detailed treatment of the larvae and life history of these animals will occupy a separate work.

Franclemont wrote in his description of race benjamini (1955: 46): “Similar to the typical race, but the markings are a little less distinct, and the color is generally duller. All the specimens are a rather uniform tannish brown with a contrasting purplish gray terminal area.” He described the male genitalia as “somewhat larger than the typical race” and the female as having the “ductus bursae more heavily sclerotized; general characteristics similar to those of forbesi.” The paratypes of benjamini comprise specimens from East New York [Brooklyn], NY; Connecticut; and Martha’s Vineyard, MA. Upon close examination of these and other specimens, including a series of Massachusetts specimens and the holotypes of A. crumbi, A. nefascia, A. forbesi, and A. reedi, we noticed consistent, if subtle, differences between eastern specimens and typical A. nefascia. In addition to those differences in coloration noted in Franclemont’s (1955) description, we focus on the configuration of spines on the vesica as well as characters adduced by Franclemont (wing pattern and sclerotization of the female ductus), Buckett (number of signa on the bursa copulatrix), and Lafontaine (secondary diverticulum in the male vesica).

Buckett (1968c) characterized A. crumbi in a different species group than A. nefascia [sensu (Buckett 1968b), not A. nefascia (Smith, 1908)]. Although he accurately described the terminal area of the benjamini forewing as fading inwardly, he characterized both A. nefascia and A. crumbi as lacking heavily sclerotized ductus bursae and, more importantly, described the bursae of A. crumbi and benjamini as having two signa, as in Abagrotis alternata (Grote, 1864) (Buckett 1968c: 4) versus a single signum on the bursa of A. nefascia (sensu Buckett 1968b: 19). We failed to observe more than one signum on any Massachusetts specimen; most of those we examined from the type series of A. crumbi. All the more recently collected western A. nefascia possessed two.

As Lafontaine suggested, subtle features of wing pattern may be unreliable indicators of phylogenetic affinity in Abagrotis. That said, both Franclemont and Lafontaine acknowledged trends in coloration and shading among eastern A. nefascia that may serve to differentiate most specimens from their western counterparts. Eastern specimens range in forewing ground color from a gray brown to brick red; we have not seen any with a dark-maroon forewing ground color.

In addition to exhibiting the secondary diverticulum putatively diagnostic for A. nefascia (Lafontaine, 1998: 221), the scobinate plate on the vesicae of eastern specimens tends to be less evenly denticled. The basal sclerotized patch consists of a more heteroideous series of teeth, with two or three very prominent and 5–8 much smaller (Figs 19–23). Specimens with the most conspicuous teeth tend to have them arranged in a nearly co-planar fashion, almost giving them the appearance of a circular saw blade. Western specimens typically have more than 12 teeth comprising a more graduate series of size, and always arranged in multiple rows.

Finally, we observe that the presence of signa on the female bursa has been interpreted inconsistently in this group. Buckett (1969b: 19) specifically recorded A. nefascia as “possessing a single signa [sic]” and A. crumbi (as a member of the A. alternata species group; Buckett 1969c: 4) as “possessing two signae [sic].” Lafontaine (1998: 205) described Abagrotis females as having the “bursa lacking signa...or with two signa.” In the material we examined, we noted the presence of two signa, although one highly reduced, in both A. nefascia and A. crumbi, contraBuckett (1969c) (Fig. 24). However, we failed to observe a second signum, reduced or otherwise, on the bursae of ten female A. benjamini examined.

Interspecific variation in COI sequences in Abagrotis is atypically low for a noctuid genus. In the analyses of Zahiri et al. (2014), 29 of 248 genera examined account for 101 identified examples of interspecifically shared haplotypes, with Abagrotis accounting for more such incidents (eight) than any other genus examined except Euxoa (17). Therefore intraspecific divergences of less than 1% are typically considered more significant within species of Abagrotis. Our Massachusetts specimens, however, shared sequences with specimens from New Brunswick, and these collectively differed from a cluster of reliably determined specimens comprising several species with shared haplotypes, including A. nefascia.

We typically eschew the use of distance-based analyses of character data as rigorous arbiters of either phylogenetic relationship or species diagnosis (Goldstein and DeSalle 2011). Such interpretations emulsify potential diagnostic data by masking the visualization of character state distributions on trees, and contradict principles through which statements of monophyly can be legitimately tested, especially for data as minimal as the short sequence conventionally used in DNA barcoding efforts. Instead, we view distance analyses of DNA barcodes as provisional estimates of potential species boundaries, and employ a more transparent cladistic analysis for the limited purpose of summarizing and visualizing character state distributions. In this case, the clustering of eastern “nefascia” from Massachusetts and New Brunswick is corroborated by a unique combination of thirteen character state changes, greater than the combined interspecific variation among specimens of seven species in the adjacent cluster. We note an analogous cluster of specimens from British Columbia determined as A. nefascia.

In Aquinnah, high numbers of adult S. riparia were observed, routinely exceeding 200 specimens in a single light trap, the highest numbers appearing between 21 June and 5 July. Based on our capture dates, the flight season of S. riparia is somewhat protracted, extending from mid-June through early August. Females retained for obtaining eggs oviposited with varying success, most readily when presented with honey water and in the presence of beach plum foliage. Eggs proved difficult to overwinter without succumbing to desiccation or mold but efforts involving chilled hydration and aureomycin-laden artificial diet are ongoing.

Other species appearing in unusually high numbers included A. benjamini, the erebids Argyrostrotis anilis (Drury, 1773), Catocala herodias Strecker, 1876, and Drasteria graphica Hübner, 1818; and the noctuids Schinia spinosae (Guenée, 1852) and Eucoptocnemis fimbriaris (Guenée, 1852). Species apparently associated with beach plum included regular occurrences of Wild Cherry Sphinx (Sphinx drupiferarum J.E. Smith, 1797), which has declined in the Northeast (Wagner, 2012), and now appears confined to coastal areas (Goldstein et al. 2015). During the course of our work we observed other widespread but often scarce Prunus-feeding moths that have declined. In particular, both adult Hyalophora cecropia (Linnaeus, 1758) and cocoons on beach plum were observed frequently. Of equal ecological interest were the high numbers of moths associated with host plants other than Prunus or other signature dune species, and in particular species associated with scrub oak such as Catocala herodias and Cicinnus melsheimeri (Harris, 1841) (Mimallonidae).

Finally, we collected at least four species of Abagrotis in addition to A. benjamini, namely A. alternata, A. brunneipennis (Grote, 1875), A. cupida (Grote, 1865), and A. magnicupida Lafontaine, 1998, some of which have since been reared as wild-caught larvae from beach plum. Abagrotis benjamini adults first appeared in late June, apparently aestivating and appearing in highest numbers later in the season, primarily in August but extending through September.

Initial efforts to locate larvae on beach plum at Aquinnah in 2012 and 2013 were less immediately successful due to a combination of seasonal conditions and timing. Larvae of Sympistis riparia were collected from beach plum and reared to adults at a mainland site in Dartmouth, MA on 16 and 22 May 2013 (Fig. 7a–c) and at Aquinnah on 25 May 2015 and 18–26 May 2016 (Fig. 8a–d). At both sites, S. riparia larvae were found alongside those of what turned out to be Abagrotis benjamini (Fig. 9a–d). These observations suggest a potential association of both species with beach plum, consistent with known rosaceous hosts recorded from species in both genera. Larvae were not at first found in numbers, when the beach plum was in full but early flower, suggesting a very early hatching (S. riparia) or emergence (A. benjamini). A more intensive survey over seven nights in May 2016, timed to coincide with the first flowering of beach plum, documented over thirty late-instar larvae of each species. Most larvae were observed feeding on flowers and always on low growth (<1m) after 2100h. In addition to rearing larvae of S. riparia and A. benjamini to adults (Figs 10, 11), we documented larvae of several other noctuines actively feeding on beach plum, including Abagrotis alternata and Abagrotis magnicupida.

Larvae of both Abagrotis benjamini and Sympistis riparia are subterranean by day, and we were impressed in particular by the ease with which S. riparia larvae move through sand. As has been observed of other Sympistis, larvae of S. riparia construct an underground cell within which to pupate. With S. riparia, the cell is approximately 3 cm below the surface of the sand, and lined with a thin layer of silk. For this reason, the precise time of pupation was difficult to pinpoint. But, based on our extraction of some larvae and, in the case of Abagrotis, their successful pupation in paper towels, the pupal period for S. riparia was under 20 days; that of A. benjamini was somewhat longer, between 25 and 33 days.

The maturity of the S. riparia caterpillars at the time of their observation in the field was surprising given the brief time their host had been in flower (no more than 10 days). Some Sympistis (e.g., S. forbesi Zacharczenko & Wagner, 2014, S. piffardi (Walker, 1862)) are known to over-winter as eggs (McCabe 1985; Zacharczenko et al. 2014), others (e.g. S. perscripta (Guenée, 1852)) as pupae (Wagner et al. 2011), but none as larvae. We suspect that larvae of S. riparia emerge prior to the expansion and initially feed on buds, the most available meristematic tissues.

Our observations of mature Abagrotis larvae, on the other hand, are consistent with those of Rings (1971, 1972a,b), Crumb (1956), and Lafontaine (1998: 206) to the effect that “most, if not all, species [of Abagrotis] pass the winter as partially grown larvae.” There had been some confusion concerning the diapause of Abagrotis in that Buckett (1968a) characterized its members as overwintering as eggs. What makes this life history especially striking in the case of Atlantic Coast dune-dwelling caterpillars, and those at this study site in particular, is the highly disturbance-prone nature of the habitat and the intensity with which it has been visited by recent storms (Fig. 12).

Material examined.

Repository abbreviations.

The following abbreviations refer to collections from which specimen material forms the basis of this work:

AMNHAmerican Museum of Natural History, New York, USA

CUICCornell University Insect Collection, Ithaca, New York, USA

UCDCBohart Museum, University of California, Davis, California, USA

USNMNational Museum of Natural History [formerly, United States National Museum], Washington, District of Columbia, USA

Beyond the history of nomenclatural confusion surrounding Abagrotis nefascia and its synonyms, complicating matters further is the difficulty in locating type material, some of which is incompletely labeled or lost, including the genitalic preparation of the benjamini holotype. Notwithstanding the apparently mixed type series of A. nefascia (Lafontaine 1998: 234), both the holotypes of A. crumbi and A. crumbi benjamini are unaccompanied by USNM slide numbers, and since the holotype slide of A. crumbi was not so labeled, it is unlikely that the A. benjamini slide (wherever it resides) was either. The dissection described for the holotype in Franclemont’s original description (JGF 1203) appears lost, although a slide bearing the same data with an adjacent number (JGF 1202) was located.

Meanwhile, Benjamin’s dissection (FHB 850), which represents the holotype slide of A. crumbi as per Franclemont’s (1955) description, was originally labeled “Paratype” at the time of its preparation in 1934, more than twenty years prior, while another of Franclemont’s paratypes (FHB 843) was labeled “Holotype” by Benjamin; the name crumbi having apparently originated with him, based on the type material originally reared by S.E. Crumb in 1933. Two more recent preparations made by Franclemont of male A. crumbi, dated 14 May 1990 (JGF 7627 and 7628), are unaccompanied by specimens either at USNM or CUIC.

However, most of Franclemont’s paratype material for A. crumbi (24 of 27 specimens) and A. benjamini (5 of 8 specimens) is intact, as are series of adults and accompanying dissections of Massachusetts Abagrotis “nefascia” collected by Jones and Kimball on Martha’s Vineyard and Nantucket, the dissections made subsequently by A.E. Brower.

Type material.

Abagrotis nefascia. (Figs 13a, b, 18). Lectotype (♂, AMNH): New Mexico: Ft Wingate NM VII.21; J.B. Smith Collection Rutgers; Rhynchagrotis nefascia ♂ type Sm.; Lectotype Rhynchagrotis nefascia Sm. By J.S. Buckett; ♂ Genitalia mounted on slide, F. H. R. no. 12,106; Slide labels: ♂ Genitalia Rhynchagrotis nefascia J. B. Smith Ft. Wingate, N.M. VII.27; No. 12,106 LECTOTYPE BALSAM Mounted XI.18.1963 Fred H. Rindge

Abagrotis crumbi. (Figs 15a, b, 20). Holotype (♂, USNM): Washington: White Swan VI-5-33; White Swan No. 3; Collector S.E. Crumb; 53; 8; ♂ gen. 850 7 Feb. 34 FHB; HOLOTYPE Abagrotis crumbi J.G. Franclemont. Paratypes (12 of 14 ♂, 12 of 13 ♀ designated by Franclemont): Washington: White Swan (5♂, 6♀); Ellensburg (6♂, 3♀); Yakima (1♂, 1♀); Tieton (1♀); Cashmere (1♀)

Abagrotis reedi. Holotype (♂, UCDC): Holotype ♂ California: Abagrotis reedi J.S. Buckett; Tecate Pk Cal San Diego Co VII-21-1963; Bill Reed Collector; 226

Abagrotis forbesi. Holotype (♂, USNM) Utah: Lampra forbesi Benj. Holotype ♂; T Spalding IX-19-7 Stockton, Ut; Barnes Collection; Slide No. 20640.

Other material examined.

Abagrotis nefascia. USA: COLORADO (2♀): Oak Creek Cany Col VII.12; Col. Jacob Doll.; USNM Dissection 148058♀; Glenwood Spgs Colo.; July 24–30; Barnes Collection; USNM Dissection #148049♀. WASHINGTON (2♂) (Figs 14a, b, 19a, b): Yakima Co. 4 mi SW Tampico 2 July 2007 BL trap, P.J. Landoldt; Barcode of Life DNA voucher specimen SampleID CCDB-28975-B03 BOLD Proc ID LNAUU1440-15; USNM Dissection 148023; USNMENT01203911; Barcode of Life DNA voucher specimen SampleID CCDB-28975-E03 BOLD Proc ID LNAUU1476-15; USNM Dissection 148024; USNMENT01203941; WYOMING (1♀): Converse Co. 0.3 mi S of Glenrock Mormon Canyon Rd. 7 Aug. 2002 42 51.43'N, 106 51.75'W MG Pogue at UV trap; Barcode of Life DNA voucher specimen SampleID CCDB-28975-E02 BOLD Proc ID LNAUU1475-15; USNM Dissection 148025♀; USNMENT01203940. UTAH (2♀): Daggett Co. Brown’s Park on Green River 1 August 2005 J.D. Hooper Coll.; UT: Daggett Co. elev. 5490' 40.54' 04.2"N, 109.08' 40.3"W; USNMENT01279055; Uintah Co. 11mi NNE of Vernal 1 Sept. 1997 @ light J.D. Hooper Coll.; Unitah Co. elev. 5850’ 40 35’ 38.8'N, 109 25’ 37.1'W; USNMENT01279080.

Abagrotis magnicupida. USA: MASSACHUSETTS (2♂, 2♀): West Tisbury, Manuel F. Correllus State Forest, Willow Tree Bottom, 20 July 2004; P.Z. Goldstein leg. (1♂, 3♀); USNM Dissection #148007♀; 148008♂.

Abagrotis brunneipennis. USA: MASSACHUSETTS: Edgartown, Manuel F. Correllus State Forest, bathtup deerstand 41°23.770'N, 70°35.550'W, 30 June 2011 @ UV trap P.Z. Goldstein (1♂).

Abagrotis alternata. USA: MASSSACHUSETTS (1♂, 2♀): West Tisbury, Manuel F. Correllus State Forest, Willow Tree Bottom 22 August 2000 (1♂, 1♀); USNM Dissection #148009♂; 148010♀; 20 VII 2004; USNM Dissection #148011♀.