Adult morphology of all eastern North American species in the glaucus-complex can be deceivingly similar, and any single morphological character should not be relied upon for identification. Most similar to P. solstitius are P. glaucus, P. canadensis and potentially P. bjorkae, another new species in the glaucus-complex proposed in 2024 (Pavulaan 2024). Given its recency, the taxonomic status of P. bjorkae has not yet been scrutinized by the scientific community, but it is necessary to do so here. For the reasons detailed below the recognition and diagnosis of P. bjorkae is currently problematic, although based on the spring flight period and comparison of the figures in the original description (Pavulaan 2024), it is certain the name does not apply to MST.

The justification for treating P. bjorkae as a distinct species hinges on recognition of three distinct, partially sympatric, spring-flying taxa, recognized by adult phenotypes (P. glaucus, P. “ near canadensis,” P. bjorkae) which correlate with slightly different flight periods (Pavulaan 2024). No diagnostic differences in immature stages, biology, larval hosts, or molecular markers of P. bjorkae have been documented to date (Pavulaan 2024), nor is there evidence in previous research that might hint at the existence of such (e.g., Ording et al. 2010; Kunte et al. 2011). Using seasonal adult abundance peaks combined across the glaucus-complex, flight phenologies for taxa present within the range of P. bjorkae are attributed to spring (P. glaucus, P. canadensis, and P. bjorkae), summer (midsummer swallowtail), and late summer (second-generation P. glaucus) (Pavulaan 2024: figs 3–5). During spring (May through June), P. bjorkae flies in “late spring,” versus “early spring” for P. glaucus and P. canadensis. However, only a single spring abundance peak is evident and attributed to P. bjorkae, whereas neither P. glaucus nor P. canadensis peaks are distinguishable due to the relative scarcity of observations for these species (Pavulaan 2024: 7, figs 3, 4). No additional data are provided to define late- versus early spring, leaving it unclear to what extent the phenology of P. bjorkae differs. Life history data that could corroborate such a difference are currently lacking.

The differential diagnosis of P. bjorkae is based largely on differences in wing pattern and shape, especially of the female (Table 1). Males are described as intermediate between P. glaucus and P. canadensis; comparative differences are given compared to P. appalachiensis and P. canadensis, but not P. glaucus (Pavulaan 2024: 16). Without an indication of sample size and a full description of male and female morphology, it is currently difficult to gauge intra- versus interspecific variation. Lastly, P. bjorkae is stated to be larger than spring P. glaucus and P. canadensis, but conflicting information on p. 9 states that P. glaucus is the largest species in the study region. No size measurements specific to male or female are given for P. bjorkae (including the holotype), nor is it possible to infer size of specimens from figures since scale bars are not given; size as a diagnostic trait for P. bjorkae therefore remains undefined.

The adult phenotype of P. bjorkae is very similar to that of P. canadensis and P. glaucus, so attributing phenotypic variation to three different putative taxa requires careful assessment. A potential additional source of phenotypic variation which remains unstudied stems from seasonal polymorphism in P. glaucus. Contrary to the assumption that P. glaucus is obligately bivoltine at the northern range edge (Pavulaan 2024), Ryan et al. (2016) demonstrate that it can be uni- or bivoltine depending on thermal constraints. In other words, temperature and day length experienced during the larval stage of P. glaucus dictate whether or not pupae develop directly into second generation adults, or enter winter diapause to emerge the following spring (Ryan et al. 2016). Since adult phenotype of P. glaucus is influenced by different temperature-photoperiod profiles (different spring and summer forms are well-known in P. glaucus, e.g., Pavulaan and Wright 2002), populations that comprise uni- and bivoltine cohorts would be expected to exhibit bimodal spring phenotypes (i.e., those developed from previous year’s spring versus summer adults). If proven, phenotypic variation driven by facultative voltinism in P. glaucus could account for the perception of phenotypes that are unaccounted for with existing taxonomy.

It is evident that the descriptive and diagnostic information defining P. bjorkae is currently incomplete and partially contradictory, and corroborating evidence for its distinctness as a species, outside of adult morphology, is lacking. This renders the recognition of P. bjorkae as a valid species tenuous at best. To spur further inquiry and study, we nevertheless include the known comparative phenotypic traits in Table 1.

Despite the overall similarity of P. solstitius to P. glaucus, we have found that it is possible to confidently identify the vast majority of individuals when multiple diagnostic traits are assessed. Papilio solstitius is most similar to the northernmost populations of spring generation P. glaucus, and some specimens are not distinguishable based on wing pattern alone. Papilio solstitius differs from P. glaucus in smaller overall size, greater tendency for the ventral forewing submarginal band to be band-like (broken into rounded crescents interrupted by black veins in typical P. glaucus); less scalloped outer border of the ventral hindwing submarginal band, and the absence of dark phase females (present in both P. glaucus and P. appalachiensis). The forewing outer margin is less frequently concave than in P. glaucus. Variation in these wing pattern traits often overlap with those of P. glaucus, and specimen identification requires consideration of seasonal timing and location. In P. solstitius, the tuft of setae projecting from the frons is much more prominent than in summer generation P. glaucus, where it is greatly reduced (Fig. 4); spring generation P. glaucus have similar setation to that of P. solstitius. The spring generation of P. glaucus can have some P. canadensis-like traits (Scriber 1990) that make it more difficult to differentiate from P. solstitius based on adult morphology alone. However, throughout much of the range of P. solstitius, there is no overlap with the more southern P. glaucus. Male genitalic structure is generally regarded as being homogenous among the glaucus-complex (Brower 1959; Hagen et al. 1991), but our limited sample suggests that there may be quantitative differences in the number of dorsal tines on the clasper, with P. canadensis and P. solstitius ranging from one to two spines and P. glaucus from one to three (Fig. 9).

Compared to sympatric P. canadensis populations, P. solstitius can usually be separated with confidence. It is larger with less extensive black markings, most consistently so in the narrower black border of the hindwing anal margin (Fig. 7; Table 1). The narrower margin also results in the large black V (formed by the medial line bridging to the distal part of the anal margin) appearing more U-shaped, versus more sharply V-shaped in canadensis (Fig. 7). The ground color is a slightly richer yellow tone. The body vestiture and color differ significantly between the two: the setation of P. solstitius is more sparse and shorter, particularly evident on the frons (Fig. 3), the dorsal thorax, and along vein 2A through the black anal margin band of the ventral hindwing (Fig. 7). The head and dorsal thorax are brighter yellow, as is the abdomen. The abdominal subdorsal yellow band is also wider, the male clasper is solid yellow, not interspersed with grey-black scales as in P. canadensis (Fig. 8).

Best observed on the underside of the hindwings, the anal margin black band relative to the width of the entire cell containing the band is approximately 10–40% wide in P. glaucus and 55–90% wide in P. canadensis (Scriber and Ording 2005). The band width averages greater in females than males, but the relative difference between species persists. In P. solstitius, this width ranges between approximately 30–55%. Also, on the underside of the hindwings, the lateral interface separating the basal yellow from the black submarginal region is typically somewhat straight in P. canadensis (though a common exception being in cell Rs-M₁ where the line can be bowed inward), noticeably scalloped in P. glaucus, with P. solstitius demonstrating intermediacy. The hindwing underside submarginal lunules tend toward those of P. canadensis in being more rectangular than crescentic.

Comparison of the larval morphology indicates that the color pattern of the first instar is diagnostic for P. glaucus and P. canadensis (Hagen et al. 1991; Scriber 1998). Papilio solstitius differs from P. glaucus and P. canadensis in the white dorsal banding pattern (Fig. 10). The prominent white medial saddle, comprised mostly of segments A3-A4, is present in all species. In P. canadensis, there are three additional, smaller white bands: two anterior bands formed by white pigmentation on T1 and T3, and a posterior band formed by A8. This banding pattern, with additional anterior-posterior (AP) bands, is consistent in P. canadensis. In P. glaucus, only the A3-A4 medial saddle is present, and AP bands are absent, the pigmentation on T1, T3, and A8 being dark brown. Papilio solstitius shows intermediacy and variability in the development of the AP bands. Typically, the AP bands are not as prominently white as in P. canadensis, but not completely brown as in P. glaucus. Development of the AP patterns varies and can be absent (glaucus-like) or highly developed (canadensis-like), although such variants are rare (< 10% of individuals reared). However, canadensis-like larvae never express the same intensity of white pigmentation as that species, although dark variants are essentially undistinguishable from P. glaucus. Examples of glaucus-like first instars are limited to one field-collection event on a single ash sapling (Kingston, 22.Jul.2023), where five of nine larvae were glaucus-like. Clearly, further study of larval variation is needed.

In the southern range parts, Papilio solstitius seems to prefer ovipositing on tulip tree (Liriodendron tulipifera L.) and hoptree (Ptelea trifoliata L.), like P. glaucus. Larvae can occur regularly on hoptree where it is planted as an ornamental shrub outside of the natural range (Fig. 12). North of the native ranges of both of these plants (approximately north and east of the region of Toronto, Ontario), P. solstitius feeds on Fraxinus pennsylvanica and Prunus serotina, based on wild-collected ova and larvae and observation of oviposition (Fig. 4b). Larvae demonstrate high survival rates on tulip tree, unlike P. canadensis, and also demonstrate survival on quaking aspen (Populus tremuloides Michaux), unlike P. glaucus, but at a rate lower than that of P. canadensis (Mercader et al. 2009).

Figure 12. 

Distribution of examined specimens of P. solstitius (voucher data in Suppl. material 1).

Papilio solstitius exhibits delayed post-diapause pupal development, producing a single summer flight. In Ontario, the flight period commences in late June to early July, peaking in the first half of July (Fig. 2a). Studies on the effect of temperature on pupal development show a similar phenology in New York (Ording et al. 2010). Rearing field-collected ova and larvae from the Kingston region of Ontario further confirm that P. solstitius is univoltine with obligate diapause like P. canadensis, differing from P. glaucus which is facultatively multivoltine (Scriber 2013). Notably, some lab reared pupae overwintered twice, not eclosing until the second year.

Pupae removed from cold storage to a constant temperature of ~23 °C eclosed after 30.4 +/- 5.5 days (male and female combined), or an average of 699 degree-days (DD). Papilio canadensis pupae emerged 19.4 +/- 4.2 days (p < 0.0001), or 446 DD, under the same conditions. In eastern Ontario, accumulated degree-days (above a minimum threshold of 6 °C) for these values correspond to the second week of June (446 DD) and the first week of July (699 DD) (Schmidt and Layberry 2016), precisely when peak emergences of P. canadensis and P. solstitius are occurring (Fig. 2a). Difference in post-diapause pupal emergence therefore perfectly accounts for the staggered emergence peaks between P. canadensis and P. solstitius in eastern Ontario. Male and female P. solstitius differ in the length of post-diapause development delay. On average, males required approximately 26.6 +/- 3.2 days to eclose compared with 34.2 +/- 4.2 days for females (p = 0.02; n = 10; two-tailed T-test). In the wild, this would be expected to translate to a difference in peak flight times between the sexes of approximately 15 days, which matches well with field observations (Fig. 2).

Bivoltine P. glaucus populations occur primarily to the south of the range of P. solstitius. However, P. glaucus is facultatively univoltine or bivoltine at the northern range periphery, contrary to the initial hypothesis that it is unable to switch to univoltinism and limited to regions where it can undergo two annual generations (Hagen et al. 1991). In Ohio and Michigan populations, pupae are induced to enter winter diapause when 4^th-5^th instars experience photoperiods of less than 14 hours (Ryan et al. 2017). Facultative uni- vs. bivoltinism is also demonstrated by our rearing results from the Hamilton, Ontario region, which is north of the bivoltine thermal threshold (Scriber 2013). Lab-reared larvae of spring P. glaucus on L. tulipifera developed directly into a second generation of adults, despite the rarity of naturally occurring second-flight P. glaucus here. Univoltine P. glaucus populations probably occur more widely than previously recognized and have added to the complexity of defining the taxa involved in the glaucus-complex. Indeed, this could explain the perception of two spring-flying phenotypes (Pavulaan 2024) in regions where both uni- and bivoltine P. glaucus occur: offspring developing from either spring-flight (univoltine) or summer-flight (bivoltine) parents experience differing temperature-photoperiod profiles as larvae (known to influence adult phenotype), but both cohorts emerge the following spring. In southern Ontario and the Finger Lakes region of New York, the presence of both spring and summer P. glaucus likely accounts for a longer spring abundance peak and a more protracted late summer abundance peak (Fig. 2a, c; see also Schmidt 2020: fig. 7).

Since Papilio solstitius, like its congeners, uses a range of unrelated host plants, it has a similarly broad habitat tolerance for a range of forest, forest edge and woodland habitats. Although habitats of P. solstitius and P. canadensis overlap widely, the former reaches its highest abundance in or near mesic or moist woodlands, particularly ash-dominated swamps, where ash is common. Conversely, P. canadensis is most common in drier upland habitats where trembling aspen is common.

The core range of Papilio solstitius includes eastern and southcentral Ontario, northern and central New York and adjacent Vermont, New Hampshire, and Pennsylvania (Fig. 12), encompassing a minimum land area of approximately 174 000 km² (by comparison, the range extent of P. appalachiensis is ~ 140,000 km²). In New York, P. solstitius inhabits most of the state except the southeast and greater New York City area. In Canada, P. solstitius extends from the Montréal, Québec region west to the Bruce Peninsula of Ontario, south to the Niagara region (Fig. 12; Wang 2018; Schmidt 2020; Macnaughton et al. 2020). The western limit appears to be the eastern shores of Lake Huron; we have not seen any verifiable specimens west of there. The glaucus-complex has received considerable study in the lower peninsula of Michigan and in Wisconsin, and there is no evidence of delayed flight (July) swallowtails there (Luebke et al. 1988; Stump et al. 2003).

The northern range limit of P. solstitius is easily defined since adult morphology and phenology differ considerably from P. canadensis. Furthermore, the range limit is climatically constrained since P. solstitius larval development is shifted about a month later than P. canadensis, and development must be completed before autumnal leaf abscission and frost. The current northern limit is the southern edge of the Algonquin Dome, the lower Ottawa River valley, and the southern edge of the Gatineau/Laurentide escarpment as far east as the Montréal region.

Papilio solstitius has undergone a northward range expansion of several hundred kilometers since the 1970s (Schmidt 2020), as has P. glaucus elsewhere (Scriber et al. 2014). In 2022, P. solstitius was recorded for the first time near Montebello, Québec. Continuous monitoring at this location since 1994 indicates that P. solstitius was not present prior to 2022 (P. Legault, pers. comm). Based on the climatic zones given in Scriber et al. (2014), the distribution of P. solstitius approximates the 1300–1400 degree-day (°C) climatic envelope. For context, the northern limit of bivoltine P. glaucus is ~1444 DD. The southern (warm) limit of P. canadensis appears to be slightly north of this, and is possibly limited by pupal mortality due to prolonged high summer temperatures (Kukal et al. 1991). The northern range limit of Papilio solstitius is likely determined by minimum thermal requirements, given the late seasonal phenology of a July flight period that dictates a shorter window for larval development before autumnal host plant senescence.

The southern range limits of P. solstitius are currently difficult to define owing to overlap and confusion with single- and double-brooded P. glaucus, and the uncertainty in the northern range limit of P. glaucus. Swallowtails that are morphologically consistent with P. solstitius and eclosing in the first half of July, when P. glaucus is between flights, extend south to approximately 41, 42°N to the eastern seaboard (Fig. 12). In Pennsylvania, the southern extent of P. solstitius coincides approximately with the northern limit of P. glaucus containing dark morph females (Scriber 1996), extending from Erie to just north of Pittsburgh and east to New York City. It may also extend to the Atlantic coast through Connecticut and Rhode Island based on the phenology information in Pavulaan (2024), but this warrants further study.

The occurrence of P. canadensis at the southern range edge, near that of P. solstitius, may be more limited than depicted in some range maps (e.g., Pavulaan and Wright 2002; Cech and Tudor 2005; Monroe and Wright 2017). Our examination of putative P. canadensis photos from New York and Pennsylvania indicate that most are spring flight P. glaucus; CJD has been unable to verify the presence of typical P. canadensis in New York state south of the Adirondacks. It is possible and indeed expected that P. canadensis is undergoing a northward range contraction with warming climates (Scriber 2013), but this remains unexamined. In the Finger Lakes region of New York, members of the glaucus-complex can be observed continuously from mid-May to early September (Fig. 2b). In this region, a pale canadensis-like phenotype emerges first, followed by a tiger swallowtail in late May which has historically been referred to as “spring form” P. glaucus, and then finally P. solstitius in late June to July, and possibly a partial second flight of P. glaucus in August (although not all taxa are sympatric everywhere).

Both regions of COI recover the same general relationships between members of the P. glaucus group, including P. multicaudata Kirby, 1884, P. eurymedon Lucas, 1852, P. rutulus Lucas, 1852, and the glaucus-complex clade of P. glaucus, P. canadensis, and hybrid taxa (P. appalachiensis, P. solstitius, etc.) (Figs 13, 14). Within the latter, P. glaucus and P. canadensis almost form reciprocally monophyletic clades in both COI5 and COI3, but in each gene, a handful of specimens fall in the opposing clade (marked with asterisks in Figs 13, 14), and P. appalachiensis falls throughout the P. glaucus clades in both genes. Papilio solstitius clusters within the P. canadensis clade, as does a handful of P. glaucus. Notably, there are few nodes with strong branch support within this clade of P. glaucus/P. canadensis/P. appalachiensis/P. solstitius, indicating close genetic similarity between all of these entities in their mitochondrial genomes. Excluding specimens with missing data in the 5’ or 3’ ends of their sequences, pairwise sequence identity for haplotypes in this P. glaucus/P. canadensis clade were > 98% for COI3 and > 97.5% for COI5.

Figure 13. 

Maximum likelihood tree for COI5. Specimens are labeled with a species epithet determination and NCBI or BOLD accession numbers. Papilio solstitius samples indicated in blue as “MST.” Specimens with asterisks indicate those that fell outside of their typical respective clade. Grey circles indicate strong node support (> 0.95 ufBS and > 0.8 SH-aLRT). All outgroup branch lengths have been edited for space.

Figure 14. 

Maximum likelihood tree for COI3. Specimens are labeled with a species epithet determination, state/province, or additional unique identifier, and NCBI accession numbers. Papilio solstitius samples indicated in blue as “MST.” Specimens with asterisks indicate those that fell outside of their typical respective clade. Grey circles indicate strong node support (> 0.95 ufBS and > 0.8 SH-aLRT). All outgroup branch lengths have been edited for space and dotted grey line is a visual link between independently shown parts of the tree.

We re-evaluated identification of specimens sequenced in Vernygora et al. (2022) and conclude that specimens noted as “intermediate” therein are mostly P. glaucus, but one is P. solstitius (samples annotated with asterisks in Fig. 15). In their SNP-based phylogeny (remade in Fig. 15), these specimens form a paraphyletic grade between typical (and more geographically distant) P. canadensis and P. glaucus; Papilio appalachiensis also falls out in this grade, and as with COI, only a handful of nodes within this broad clade were strongly supported and many of these specimens appeared admixed in Vernygora et al.’s population genetics-oriented analyses.

Figure 15. 

Majority rule consensus tree generated from 3,733 SNPs from Vernygora et al. (2022). Specimens are labeled with a species epithet determination, state/province, and unique identifier, and specimens with asterisks indicate those called “intermediates” in the original publication, including one we identify now as P. solstitius (“MST” in blue). Grey circles indicate strong node support (> 0.9 posterior probability support). All outgroup branch lengths have been edited for space.