North American Papilio canadensis and P. glaucus (Lepidoptera: Papilionidae, these Papilio = Pterourus) have previously been described as having allopatric distributions separated by a narrow hybrid zone running from Minnesota to southern New England, and southward in the Appalachian Mountains (possibly to northern Georgia). Recent patterns of hybridization and introgression suggest a more complex interaction between the two, possibly even resulting in the formation of a new species (Pterourus appalachiensis Pavulaan & Wright, 2002). Recently, extensive northward interspecific introgression of P. glaucus-diagnostic traits has been observed in the hybrid zone. These include wing bands and other color patterns, the ability to feed on tulip tree leaves, and Hk-100 allozymes; all are autosomally encoded. However, there has been little northward introgression of certain other P. glaucus traits (such as facultative diapause and bivoltinism, and Ldh-100 allozymes, both X-linked; and the Y-linked melanic mimicry gene in females). Interspecific recombination of the X-chromosome has evidently occurred, as shown by discordant patterns of X-linked markers. The P. glaucus X-linked Pgd-100 and Pgd-50 alleles have introgressed 200–400 km north of the historical hybrid zone, yet the P. glaucus X-linked Ldh-100 allele has not. The allele frequency shift for both genes is more closely related to the ‘thermal landscape’ (i.e., accumulated degree-days above a developmental base threshold of 50 °F (=10 °C)) than to latitude. Delayed post-diapause eclosion of cohorts within the hybrid zone, e.g., the New York/Vermont border area, has produced a natural ‘false-second generation’ flight (a hybrid swarm of synchronous males and females, where 2300–2700 °F degree-days have accumulated each year since 1998) that is reproductively isolated from flights of both parental species. Moreover, the newly described P. appalachiensis exhibits a unique combination of traits. These include obligate diapause, a univoltine habit, and the Ldh-80 or Ldh-40 alleles (as for P. canadensis), the Pgd-100 or Pgd-50 alleles (as for P. glaucus), and a delayed ‘false-second generation’ reproductive flight period (as observed in the hybrid zone). Since 2001, a rare allele or ‘hybrizyme’ (Ldh-20) has appeared in this false second generation at high frequencies (40–50%). We hypothesize that strong selection against the facultative diapause (od-)trait (and the linked Ldh-100 allele) in regions with 2800 °F degree-days or less, and divergent selection in favor of Pgd-100 (or a closely linked trait) combined with allochronic reproductive isolation, has resulted in recombinational, parapatric, hybrid speciation. There is no evidence at present that host-plant shifts or changes in sex pheromones have driven this process, in contrast to many other speciation events in the Lepidoptera.
Evolutionary change can occur rapidly, and may be critical to the survival or extinction of a species ‘at the edge’, especially following any drastic climate change. Intraspecific genetic variation, local meta-population dynamics, chromosomal recombinations, and historical gene flow will reflect the impacts of global climate change on species or communities (Scriber, 2002a,b; Agrawal, 2003; Ashley et al., 2003; Parmesan & Yohe, 2003; Root et al., 2003). The ecological and evolutionary processes at expanding range margins may change very quickly (Thomas et al., 2001). There is an increased realization that not all hybrids are unfit, ‘unpure’, and undesirable for sustaining biodiversity (Whitham et al., 1991; Arnold, 1997). That hybrid genotypes may become adapted to new habitats and quickly become established species (Rieseberg et al., 2003) stands in contrast to the view that mutations and genetic adaptations over many thousands of years may be required for speciation (with hybridization as an incidental result). While such recombinational hybrid speciation is feasible, it has been very rare (Coyne & Orr, 2004).
Here we describe very recent genetic divergence and allochronic reproductive isolation without host specialization or pheromones for Papilio glaucus, P. canadensis, and P. appalachiensis (‘hybrid speciation’). Both glaucus and canadensis are polyphagous, and appalachiensis appears to have hybrid-like traits (able to use both Salicaceae and Magnoliaceae) as well as the common hosts such as cherries (Rosaceae). This may represent the rare, but feasible parapatric recombinant hybrid speciation in animals.
Historical hybrid zone borders have changed geographically with climate warming
The eastern tiger swallowtail butterflies from Florida to Alaska were once thought to belong to a single species with subspecies P. g. canadensis and P. g. glaucus. Hagen et al. (1991) examined numerous species diagnostic traits for these taxa, and described them as distinct species, with a sharply defined and narrow hybrid zone between latitudes near 42–43 °N, from Minnesota, Wisconsin, Michigan, and New York State into southern New England (Scriber, 1996a). Recent extensive genetic introgression (1998–2004) for some, but not all, diagnostic traits (Table 1; Figure 1) has unraveled these once-concordant, coincidental trait clines (Scriber, 2002a; Scriber et al., 2003). Recent climate warming has revealed the processes involved in divergent selection on X-chromosome loci, reproductive isolation, incipient speciation, and the likely origins of a recently described species, P. appalachiensis. It is likely that it represents ‘hybrid animal speciation’. To our knowledge, the direct impacts of global warming in the origin of a new animal species through enhancement of interspecific hybrid introgression, X-chromosome recombination, and strong divergent selection for a purported hybrid species by thermal constraints on voltinism has never been reported.
Table 1. Average °F degree-days accumulations (above a base 50 °F) at key locations near the historical Papilio hybrid zone. The means (± SE) for the 35-year period 1963–97 are presented for comparison with the 6-year period 1992–97 and the most recent 6-year period (from 1998 to 2003). Data from Zedex, Inc. and J.M. Scriber (unpubl.)
1963−1997 (35 years)
1992−1997 (6 years)
1998−2003 (6 years)
Geographic locations relative to the hybrid zone are indicated in Figure 3.
Field collection and pupal diapause termination studies
Offspring from the early flight females (May–June = ‘EF’) and from the later flight (‘false-second generation’ mid-July = ‘LF’), both from the Battenkill River Valley area at the New York/Vermont (NY/VT) border were collected as pupae from field-rearing in sleeved tree branches of wild black cherry (Prunus serotina Ehrh.). The eggs and larvae obtained inside these sleeved branches were from 25–30 EF females and 15–18 LF females. The resulting pupae were collected in mid-September 2002 and stored in darkness at 3–5 °C under controlled environment conditions until the commencement of the diapause termination experiments in 2003. Pupae were removed from winter diapause chambers and randomly assigned to three controlled environment chambers (18 °C, 22 °C, and 26 °C, all under a L18:D6 photoperiod) to determine the length of time until adult emergence. All P. glaucus pupae field-reared from females captured in low elevations of Lancaster Co. in south-eastern Pennsylvania in 2002 were used for comparison.
Oviposition bioassays and larval rearing
Field-collected (or hand-paired EF and LF virgin) females were individually set up in a rotating multi-choice oviposition arena in front of a bank of 60 W lights where they could select individual leaves of different plant species (for details, see Scriber, 1993). Eggs were collected daily and counted according to the host species selected for oviposition, and then moved into 150 mm diameter plastic Petri dishes in a controlled environment chamber (L18:D6 photoperiod, at 25 °C). Females were fed a 20% honey water solution daily and returned to the arena for continuing oviposition. Total egg production and oviposition preferences over the lifetime of individual females were determined.
Neonate larvae were randomly and gently allocated to three host plants (using a fine camel hair brush): tulip tree (Liriodendron tulipifera; Magnoliaceae) which was the favorite of P. glaucus; quaking aspen (Populus tremuloides; Salicaceae), which was the favorite of P. canadensis; or wild black cherry (Prunus serotina; Rosaceae), the common host plant of both species. Leaf petioles were supported in rubber-capped water-filled vials to maintain turgor and the leaves were changed every 2–4 days as needed until the larvae reached the first molting stage or died (neonate survival). Subsequent larval rearing to pupation was done under controlled environmental conditions at 25 °C with long days (L18:D6 photoperiod). Most of the larvae (>90%) were reared to pupation on the common host, wild black cherry, after neonate survival assays of first instar survival. Some were reared on aspen or tulip tree leaves.
Determining local seasonal thermal unit accumulations (growing degree-days)
The thermal unit accumulations (in Fahrenheit degree-days = 9/5 C degree days) during each growing season were calculated (by Zedex, Inc., Bellefonte, PA, USA) for more than 2000 geographical weather sites across the north-eastern USA, using the mean daily temperatures above a base developmental threshold of 50 °F for P. glaucus and P. canadensis. The developmental threshold was calculated as the inverse of the time to complete development when plotted against rearing temperatures (Scriber & Lederhouse, 1983; Logan & Powell, 2001). The 30-year average growing degree-day accumulations (1950–1980) have identified geographic areas of historical relevance for the P. glaucus and P. canadensis hybrid zone, as well as for other insects where the general transition in voltinism patterns from univoltine/bivoltine occur; Scriber & Hainze, 1987). Historical degree-day accumulations above 50 °F for the north-eastern USA collected by Zedex, were run in a GIS spatial program with interpolation to 1 km2 and presented using a color coding for ‘thermal distances’ to reflect isotherms at 100 °F degree-day intervals between 2300 and 2900 (and at 500 °F degree-day intervals to the north and south of this critical thermal transition zone; see Figure 1). The precise distribution of the seasonal thermal unit accumulations (1 March−31 October) for individual years (or decades) illustrate the ‘thermal landscape distances’ that ecologically incorporate influences of altitude as well as latitude across the hybrid zone. The Battenkill River Valley at the New York/Vermont border, Pendleton Co. West Virginia, and Habersham Co. Georgia all represent areas with particularly steep altitude and thermographic transitions.
Individual adult specimens were measured for forewing length (FW to the nearest mm) and the relative black band width of hindwings (in 5% units of the total distance from the inner edge to the origin of the CuA2 vein). Differences in hindwing black band widths in the anal cell have been used as a diagnostic trait between ‘canadensis’ and ‘glaucus’ for centuries (Scriber, 1982; Hagen et al., 1991). In this hindwing band, individuals of P. canadensis are generally known to range from 55 to 90%, while P. glaucus range from 10 to 40%, and known (i.e., laboratory-paired) hybrids usually range from 40 to 55% (see Scriber, 1982, 2002a).
Documenting the northern limits of Papilio glaucus and the ‘false-second generation’
The critical minimal thermal unit accumulation levels required for P. glaucus to complete a second generation (with larvae on appropriate host plants of exceptionally high nutritional quality with no seasonal decline in nutritional quality in August, and with no delays in finding mates or laying eggs) is 2700 °F degree-days (Scriber & Lederhouse, 1992). However, the realistic northern limits to bivoltine potential on the favored host species is 2800–2900 °F degree-days, which generally correlates very closely with the northernmost limits of historically documented dark morph female distribution (Scriber et al., 1996). The tulip tree detoxification abilities of the larvae (Scriber, 1982), and other species-diagnostic traits of P. glaucus, such as diagnostic allozymes (Hagen & Scriber, 1991), and morphological traits in larvae (Scriber, 1998) and adults (Luebke et al., 1988; Scriber, 1990), have shown geographic concordance of species diagnostic trait clines (Hagen et al., 1991) at the ecotone (plant transition zone) for 2–3 decades (Scriber et al., 2003). This ecotone and historical hybrid zone for Papilio has been geographically predicted by degree-day isoclines of 2300–2800 °F degree-days. However, populations at these key geographic locations have changed greatly with recent climate warming (since 1998). Despite the rapidly northward moving geographic location of the hybrid zone, ‘thermally defined hybrid zone’ locations remain at similar degree-days accumulated (2300–2800 °F degree-days), depending on which particular diagnostic trait is involved (Figure 1; Scriber, 2002a).
Historical confusion has persisted regarding the taxonomic status of P. glaucus throughout most of New York State, except in the Adirondack Mountains, the Tughill Plateau, the central Allegheny Plateau and the Catskill Mountains, where the ‘spring’ form, or ‘pure’P. canadensis, clearly predominates (all individuals with the diagnostic morphology of P. canadensis, with only a single generation, with virtually no ability to survive feeding on tulip tree, and with only yellow morph females). Throughout some suspected hybrid areas with intermediate types that are not clearly P. canadensis, the tiger swallowtail apparently has two flights in many years but not two ‘true’ generations, since there were insufficient thermal units to produce the second from the first (Shapiro, 1974; Scriber, 1975). The thermal constraints on bivoltine potential was involved in the ‘false second’ generation that was also described for Tompkins Co., New York in the 1980s by Hagen & Lederhouse (1985). Such false second generations (mid-July flights seen every year since 1999) were never seen previously in the Battenkill River Valley population from 1970 to 1997 (H. Romack & J.M. Scriber, pers. obs.). The nature of these ‘false-second’ generation flights (also lacking dark morph females) was examined from geographically extensive surveys and from intense individual trait analyses for individuals of such flights in the Battenkill River Valley region in NY/VT, which have shown warmer summers since 1998 (Figure 2; Table 1).
Changes in the thermal landscape
The total seasonal thermal unit accumulations for the past 40 years (degree-days above a base 50 °F developmental threshold) are summarized at several key locations in the eastern USA in or near the historical hybrid zone (Figure 2). Every year, from 1963 to 2003, accumulations near the juncture of the Habersham/Rabun Counties in the mountains of northern Georgia were greater than 2900 °F degree-days. Two generations for P. glaucus are usually possible anywhere the degree-days are at least 2900 °F (Scriber & Lederhouse, 1992). These Georgia counties were locations where the southernmost reference ‘type’ specimens of the new P. appalachiensis had been collected (Pavulaan & Wright, 2002). West Virginia (Pendleton Co.) appears to be at the center of P. appalachiensis records (Figure 3) and here, just below Spruce Knob, the degree-day accumulations have been generally stable (between 2400 and 2900 °F from 1963 until warmer years in 1991, 2001, and 2002). For 35 years (1963–97) the West Virginia location has averaged 2616 °F degree-days), while central New York State (Tompkins Co.) averaged 2367 °F degree-days (Table 1). The Vermont/New York border population at the Battenkill River Valley had averaged only 1934 °F degree-days from 1963 to 1997, but has averaged 2375 °F degree-days during the years 1998–2003 (Table 1). The recent 6 years have been 378–867 °F degree-days warmer across the eastern USA at these locations than the previous six years (from 1992 to 1997, Table 1, Figure 2). It is also clear that the six recent years have been warmer than the previous 6 years (as well as the previous 34 years) at the hybrid zone in Michigan and Wisconsin (Table 1).
As they had been examined annually since the early 1980s (Scriber, 1982, 1994; Scriber et al., 2003), various ‘species-diagnostic’ morphological, physiological, biochemical, behavioral, and genetic adaptations of these Papilio butterfly species had been geographically delineated across a long, narrow, parapatric distribution with little parental overlap between P. glaucus and P. canadensis. Little meaningful introgression had occurred in either direction across the hybrid zone during this 1978–97 period (Figure 3; Luebke et al., 1988; Hagen et al., 1991; Scriber, 1996b; Scriber, 2002b; Scriber et al., 2003). Recent evidence (1998–2003) of more extensive hybridization and introgression suggests a more complex interaction between the two species, and possibly even the formation of a new species, Pterourus (=Papilio) appalachiensis; Pavulaan & Wright, 2002). All of the ‘type’ specimen locations of P. appalachiensis reported by Pavulaan & Wright (2002) turn out to be counties with higher elevations and constrained thermal landscapes that would select against a second generation. All P. appalachiensis specimens are inside the hybrid zone very near areas of 2500–2800 °F degree-days annual accumulation (Figure 3; and J.M. Scriber, unpubl.).
The recently described mountain species, P. appalachiensis, has been shown to have hybrid-like traits for all morphological (and many other) species diagnostic characters (Table 2; and Pavulaan & Wright, 2002). The distinctive traits in P. appalachiensis are a ‘delayed post-diapause emergence’, a univoltine life cycle, and X-linked allozymes that are basically fixed for the canadensis-like Ldh alleles and also fixed for the glaucus-like Pgd alleles (i.e., alternative species diagnostic alleles from the two putative parental species). The geographic distribution of the new Appalachian Mountain swallowtail species (Pavulaan & Wright, 2002) has only been reported from within the historical P. glaucus/P. canadensis hybrid zone (and in the southern Appalachian Mountains in the states of North Carolina, Georgia, South Carolina and Tennessee; Figure 3).
Survival means are based on three families from West Virginia (in 2004).
Ldh-20 is a rare allele never before seen in these Papilio prior to 2001, but now was at high frequency in the ‘LF’ incipient species population (and was also seen in P. appalachiensis at low frequency; Table 3). Some hemizygous dark morph females from the West Virginia P. appalachiensis population had canadensis-like Ldh-80 or Ldh-40 allozymes, and some had Ldh-100 (as did yellow morph females; G.J. Ording & J.M. Scriber, unpubl.). Males of P. appalachiensis (essentially all with canadensis-like Ldh alleles, but glaucus-like Pgd alleles) have not yet been evaluated for dark gene enabler/suppressor status (see Scriber et al., 1996), but some of the hybrid swarm (mid-July flight) at the NY/VT border have the X-linked s– enabler gene (although no dark females had ever been collected within 150–200 km of this hybrid swarm).
One of the easiest wing traits for diagnostic separation of P. canadensis, P. glaucus, their hybrids, and the hybrid-like P. appalachiensis is the relative black band width at the proximal edge of the hindwing (Scriber, 1982, 1990; Luebke et al., 1988). Generally, the narrow 10–40% represents P. glaucus, the wide 55–90% reflects P. canadensis, and the intermediate 40–55% reflects hybrids and/or P. appalachiensis (see Tables 2 and 3). A recent narrowing trend for this morphological diagnostic trait has become obvious at many locations near the cooler (northern) side of the historical hybrid zone, including central Wisconsin, central Michigan and south-western Vermont (Scriber, 2002b, see also Table 3). Along with this apparent glaucus-like introgression of hindwing band widths there has been an increase in forewing size, tulip tree oviposition preferences of adults, and tulip tree detoxification abilities of larvae north of the historical hybrid zone, into areas thought to be occupied by basically ‘pure’P. canadensis genotypes (Scriber, 2002b; Table 4).
Table 3. The Vermont/New York (NY/VT) June flights and mid-July (false-second generation) flights compared to West Virginia (WVA) Pterourus appalachiensis and Papilio glaucus for diagnostic Pgd and Ldh allozyme frequencies and hindwing bands (all specimens are mules)
Table 4. Average (mean of family means) of neonate (first instar) survival percentage of various species/genotypes of Papilio in split-brood bioassays on three plant species, tulip tree (TT = Liriodendron tulipifera), black cherry (BC = Prunus serotina), and quaking aspen (QA = Populus tremuloides). ‘EF’ = early flight (May–June), ‘LF’ late flight (mid-July). Generally 15–150 larvae per female (family) were distributed across host plant treatments
Population near Georgia collection site of P. appalachiensis type specimen.
P.c. = P. canadensis; P.g. = P. glaucus (Hybrid mother is listed first).
Physiological detoxification abilities (tulip tree and quaking aspen)
The virtual inability of all neonate larvae of P. canadensis to survive and grow on the tulip tree (Liriodendron tulipifera, of the Magnoliaceae) had been documented as an autosomal trait (Scriber, 1986a) extensively across the eastern USA from 1980 to 1997 (Scriber, 1982, 1988; Hagen, 1990; Nitao, 1995) and basically served as a species-diagnostic trait by itself (Table 4). Natural interspecific hybridization occurring well north of the historical hybrid zone (Donovan & Scriber, 2003), and this recent movement of tulip tree detoxification genes in Michigan and Wisconsin (Scriber, 2002b) represent an extremely rapid gene flow (cf. the Heliconius hybrid zone; Blum, 2002; Dasmahapatra et al., 2002), especially when the host plant tree species distribution is not shifting northward at all noticeably.
The phenolic glycosides in Salicaceae species such as quaking aspen (P. tremuloides) have been shown to be toxic for P. glaucus (Lindroth et al., 1988). However, interspecific hybrids (and apparently P. appalachiensis; Table 3) are known to possess enzymatic detoxification capabilities for excellent survival, and capabilities for rapid growth (Scriber et al., 1989, 1999). Some recent southward movement of this trait/ability in 2002 and 2003 has been detected in southern Michigan and southern Pennsylvania, but not in 1999–2001 (Scriber, 2002b). The ability to use other hosts such as cherries (e.g., Prunus serotina) is shared by P. appalachiensis as well as P. canadensis and P. glaucus (Scriber, 1988, Table 4). Pterourus appalachiensis and the hybrid swarm at the Battenkill River (NY/VT border) seem to reflect hybrid-like abilities to detoxify both aspen, as well as tulip tree (Table 4).
Only one of 21 autosomal loci (hexokinase; Hk) and two of the five X-linked allozyme loci (Pgd and Ldh; Hagen and Scriber, 1989); provided diagnostic allozymes for P. glaucus and P. canadensis (Hagen & Scriber, 1991; Hagen et al., 1991). From the expanded analysis, including midwestern as well as new eastern populations, the steep Pgd-100 cline is evident across latitudes, and the Ldh-100 frequencies show a complex and discordant pattern with latitude (Figure 4). However, when plotted against degree-days on the thermal landscape, the Ldh-100 cline is clearly evident and appears even steeper (and at different thermal locations) than the cline for Pgd-100/50 (Figure 5). Even in areas with extensive influxes of glaucus-like traits (Pdg-100, Hk-100; tulip tree detoxification; narrowed HW band widths, larger FW, etc.) we see no Ldh-100, nor do we see dark females, nor a ‘true’ second generation. For example, populations at the Battenkill River (NY/VT border) have shown an extensive introgression of some traits since 1998 (Table 3), but we have never had reports of any dark morph females nor bivoltine populations.
At the Battenkill River population, the interspecific genetic introgression is more extensive in the ‘false-second generation’ mid-July flights, with the P. glaucus-like Pgd-100/50 alleles reaching frequencies of 39–51%. In addition, the occurrence of an apparent ‘hybrizyme’ (Ldh-20), at high frequencies in the late flight ‘hybrid swarm’ for 2002 and 2003 (Table 3), has never been recorded before in P. glaucus or P. canadensis (Hagen & Scriber, 1991). This rare hybrid-zone allele also showed up in our 2004 samples of P. appalachiensis (Table 3). Hybrizymes are rare alleles that occur only in hybrid zones, presumably as a result of: (1) increased mutation rates, (2) positive selection on the allozyme itself, or (3) ‘purifying selection’ against especially unfit multilocus hybrid genotypes which are continually purged from hybrid populations, resulting in the rare allele increasing in frequency where the selection takes place (Woodruff, 1989; Schilthuizen et al., 2001, 2004; Seehausen, 2004a,b).
Diapause regulation and voltinism
Hundreds of pupae, which were offspring from females of both the first flight (EF) and ‘false-second’ (LF) flight, had been field-reared inside large sleeve cages in the Battenkill NY/VT population in 2002. Univoltine post-diapause adults emerged in the following year (after 5 months of winter controlled-environment storage at 4 °C, in darkness) and there was essentially no temporal overlap in emergences between EF and LF individuals. Furthermore, unlike some primary hybrids (Hagen & Scriber, 1995; Scriber et al., 2002), male and female synchrony was seen within the EF or LF populations at each chamber temperature (18 °C or 26 °C; L18:D6 photoperiods; Figure 6) with slightly earlier, protandrous males (see also Nylin et al., 1993; Holzapfel & Bradshaw, 2002; Wiklund, 2003). The LF males/females were delayed (relative to the first flight emergences) an average of 12/15 days at 26 °C and 22/25 days at 18 °C. Pupal emergence times for P. glaucus in the same chambers were intermediate, but close to those for the P. canadensis-like, EF flight (Figure 6). It should also be noted that the ‘LF’ hybrid swarm sleeve-reared individuals appear to have the same autosomal glaucus/appalachiensis-like traits (longer forewings, narrower hindwing bands, and higher larval survival on tulip trees) as seen in the wild field-collected population samples from mid-July 1999–2003 (Table 3). In addition, these traits also persist in laboratory-paired and laboratory-reared families for both the early (EF) and late (LF) flights of the univoltine Battenkill populations (Table 4). These false second flights also had a significantly higher Pgd-100 frequency than the EF, but still absolutely no evidence of successful introgression of the Ldh-100 allele. Laboratory backcrosses (e.g., of a dark P. glaucus mother and a hybrid P. glaucus × P. canadensis father) produce recombinant progeny with almost all the direct-developing (non-diapause, od-) females having Ldh-100 in very close X-linkage (95%; G.J. Ording & J.M. Scriber, unpubl.).
A rapid introgression of species diagnostic traits and a rapid divergence in X-linked diagnostic alleles has been discovered inside the historical hybrid zone within 6 recent years of regional climate warming (since 1998). Geographic (latitudinal) step-cline patterns are evident for Pgd-100/50 alleles (Figure 4), but Ldh-100 allozyme frequency clines (as well as Pgd) separate much more clearly using seasonal degree-day values from the ‘thermal landscape’ (Figure 5). Two separate, non-concordant clines are evident for the species diagnostic allozyme alleles, reflecting a recombinant decoupling of these X-chromosome loci and strong divergent selection (at 2200–2500 °F degree-days for Pgd alleles and at 2600–2900 °F for Ldh alleles). The result is that the P. glaucus-like Pgd-100/50 has moved a greater geographical distance (north and up in altitude) than the P. glaucus-like Ldh-100 during those 6 years (Figure 1). Both the parental species and the newly described P. appalachiensis have parapatric geographic distributions that are uniquely distributed across the ‘thermal landscape’ for these eastern North American Papilio populations (Table 2). The false second generation hybrid swarm (putative incipient species) occupies another unique (4th) thermal niche at 2300–2550 °F degree-days along the cooler edge of the hybrid zone.
Many examples of ‘fit’ hybrids have been observed in contemporary hybrid zones (Collins, 1984; Arnold & Hodges, 1995; Arnold, 1997; Donovan, 2001; Lexer et al., 2003). Rapid chromosomal repatterning, ecological divergence, and/or spatial separation have been invoked to explain the reproductive isolation between a hybrid lineage and its parental species (Abbott, 1992; Rieseberg, 1997; Rieseberg et al., 2003). Here we have described natural univoltine ‘false second’ generation flights as delayed (but bisexually synchronized) post-diapause emergences. Similar segregation in post-diapause adult emergences have been generated in segregating genotypes of hand-paired lab backcrosses; e.g., brood no. 500, backcross of a female P. canadensis× hybrid male P. glaucus×P. canadensis (Scriber, 1990). This physiological/developmental post-diapause delay phenomenon could provide an immediate reproductive isolation of hybrids from parental gene flow (i.e., an allochronic, but sympatric mechanism of reproductive isolation).
The ecological divergence and speciation of P. appalachiensis in distinct thermal zones (that constrain them to a univoltine lifestyle) have appeared at the warmer side of the thermally delineated hybrid zone, close to P. glaucus. These locations, in the southern Appalachians of northern Georgia and the Carolinas and in the mountain ridges of West Virginia, Virginia, and Pennsylvania, represent steep elevation gradients with compressed thermal zones that would put the hybrids and parental species in closer geographic contact (and potentially facilitate hybrid interactions). The midwestern USA is generally characterized by less elevational variation and wider bands of thermal unit accumulations, and the Papilio hybrid zone (and boreal/temperate deciduous ecotone) has been wider than in the eastern USA (Scriber et al., 2003). Recent contact and extensive interspecific genetic introgression of some diagnostic traits have nonetheless been documented northward in Wisconsin and Michigan (Scriber, 2002b; Donovan & Scriber, 2003). In northern Georgia, the extremely high seasonal accumulations (even up to the mountain tops) during the past 6 years may lead to a genetic swamping of the univoltine P. appalachiensis by the bivoltine P. glaucus (Figure 2). This could extirpate the genetically unique mountain species in all areas warmed to 2900–4600 °F degree-days (Figure 2; Scriber et al., in prep).
Incipient species at the cooler side of the hybrid zone?
The most likely explanation of the origin and persistence of backcross-type hybrid swarms (i.e., individuals comprised of some interspecific introgressed diagnostic traits, but no primary hybrids that would be heterozygous for both Pgd-100 and Ldh-100) in the Battenkill River populations at the NY/VT border is allochronic isolation. At the cooler (northern) side of the hybrid zone (2200–2600 °F degree-days) there is a 2–4 week delay in the post-diapause emergences of the univoltine delayed (late flight = LF) hybrid swarm in mid-July following the normal early flight (EF) of P. canadensis types in May/June. In populations such as the Battenkill River Valley at the border of NY/VT (Washington Co./Bennington Co., respectively), there is no second generation due to seasonal thermal constraints (Scriber & Lederhouse, 1992; Scriber, 1996b, 2002b). At the warmer, southern edge of the thermally defined hybrid zone (in areas with 2600–2900 °F degree-days), the delayed post-diapause emergence of hybrid backcross-like individuals occurs 2–3 weeks after the first flight of P. glaucus and 3–4 weeks before the true second generation of P. glaucus (Pavulaan & Wright, 2002). It is such delayed false-second generations that characterize P. appalachiensis and lead to reproductive isolation from both parental species (presumably all along the hybrid zone).
Central New York State (Tompkins Co.) provides interesting historical perspectives, with populations (studied from 1970 to 1985) having a ‘false second’ generation (not derived from the first, but also univoltine with an obligate diapause; Scriber, 1975; Hagen & Lederhouse, 1985). These central NY populations were also intermediate, possessing mostly ‘hybrid-like’ traits (Scriber, 1975, 1982, 1998). The glaucus-like Pgd-100 allele frequencies had been reported at 68–76% for the 1980s and 100% for the early 1990s, while the Ldh-100 allele has remained near or at zero throughout this period to the present day (Figures 4 and 5; Hagen, 1990; Hagen & Scriber, 1991, J.M. Scriber, unpubl.). Since 1980, warming in central New York State, and Tompkins Co. in particular, may have contributed to the continued divergence in frequencies of the glaucus-like X-linked allozyme alleles, and it may be appropriate to ask if these partially isolated hybrid-like ‘late flight’ populations (Scriber, 1975, 1982) might now be classified as P. appalachiensis (Figure 5).
In addition to the different abilities to detoxify and process host species in the Salicaceae and Magnoliaceae, recent host-affiliated genetic divergence in these tiger swallowtail butterflies has also been documented for furanocoumarin detoxification on hosts such as Ptelea trifoliata (hop tree, Rutaceae; Berenbaum & Zangerl, 1997). The hop tree is the second favorite host (after the tulip tree) for P. glaucus in the eastern USA (Mercader & Scriber, 2004), and is a local favorite in central New York (Scriber, 1972). Cytochrome P450 mono-oxygenases appear to be involved in Furann coumarin detoxification (Li et al., 2002). Some gene families share a common ancestor, which over evolutionary time has been followed by subsequent differentiation into CYP6B4 and CYP6B17 groups before the speciation of P. canadensis and P. glaucus. The role of the host plant preferences of adults, and the detoxification abilities of larvae remains poorly understood for P. appalachiensis, yet it is possible that ancestral genes for processing the Lauraceae and Rutaceae may have been differentially retained in populations across the thermal landscape of the hybrid zone. However, while various local host preferences and host-specific larval adaptations have been described (Scriber, 1986b, 1996a; Bossart, 1998), there is likely to be little host race formation in these very polyphagous Papilio. In addition, there are almost certainly no host-affiliated host mating behaviors (Lederhouse, 1995; Deering & Scriber, 2002; Stump & Scriber, 2004), as may be involved in other potential sympatric speciation examples such as Rhagoletis (Bush, 1994).
This asynchrony for the ‘false second’ generation of mid-July hybrid swarms observed in the Battenkill River Valley at the NY/VT border since 1998 (Scriber & Ording, in prep; Table 2) may have resulted in partial reproductive isolation from individuals of the P. canadensis parental type (and incipient species status on the cool side of the hybrid zone). Similar delayed post-diapause emergences resulted in isolation from both generations of the P. glaucus parental species on the warmer side of the thermally defined hybrid zone in the mountains of West Virginia/Pennsylvania (Figures 1 and 2). This scenario appears to reflect sympatric/parapatric hybrid speciation, and provides an excellent system for examining processes of speciation ‘in the act’ (Figures 4 and 5; Via, 2001). With recent climate warming, the frequency divergence of the X-linked Pgd-100 and Ldh-100 of the Battenkill River population parallels that seen earlier at similar thermal constraints in central New York (Tompkins Co.) in the 1970–80s (Table 2; Figures 4 and 5).
Ecological selection and voltinism
Sometimes, because of phylogenetic and genetic architecture constraints, various trade-offs among components of life history traits are required within and between species (Danks, 1994; Chippendale et al., 1996; Zonneveld, 1996; Nylin & Gotthard, 1998; Logan & Powell, 2001; Roff, 2002). Diapause is a life-history trait that is often under very strong selection in insects (Tauber et al., 1986; Danks, 1994). Certain mortality of any second generation attempts for P. glaucus north of the hybrid zone results from insufficient time for the second generation larvae to complete their development to diapausing pupal stages (even with phenotypic plasticity and various genetically based adaptations; Kukal et al., 1991; Ayres & Scriber, 1994; Scriber, 2002a). However, selection for obligate diapause after only one generation south of the hybrid zone might also be ecologically costly, since foregoing diapause would allow the spring generation to leave more descendents in bivoltine populations. This voltinism dichotomy, which is regulated by X-linked genes, may have led to the divergence in P. canadensis and P. glaucus, and further contributed to the emergence of a third ‘hybrid’ species, P. appalachiensis.
The interspecific hybrid introgression of certain species’ diagnostic traits, between P. glaucus and P. canadensis, has generated recombinant X-chromosomes, as shown by the discordant patterns of X-linked markers (geographically, and also across the thermal landscape). The surviving genotypes in the hybrid swarm of synchronous males and females at the New York State and Vermont border are univoltine (with the X-linked od+ gene for diapause regulation), with hybrid-like detoxification abilities and morphology, but lacking the Ldh-100 alleles and without the Y-linked b+ melanism gene (of dark morph females; Scriber et al., 1996). It is clear that these ‘false second generation’ mid-July flights are delayed enough in their post-diapause adult eclosion from the first flight that they represent a partially (or nearly totally) isolated population (Figure 6). Recent detection and the high frequencies of the hybrizyme Ldh-20 in the ‘false second’ flight of the hybrid swarm of the Battenkill River Valley, and its detection in P. appalachiensis, will be examined in relation to potential functional roles in delayed post-diapause adult eclosion and relationships with other X-linked recombinants. Hybrizymes, which may be single base pair substitutions (Bradley et al., 1993; Johns & Somero, 2004), are assumed likely to be transient in nature due to recombination eventually breaking up the linkages (Barton & Hewitt, 1985; Hoffman & Brown, 1995; Schilthuizen et al., 2001). However, the fixation of hybrizymes in conjunction with recombinant hybrid speciation may persist as novel genotypes (Arnold, 1997; Coyne & Orr, 2004; Schilthuizen et al., 2001 and 2004; Seehausen, 2004a,b), as we suspect to be the recent case in our Papilio hybrid zone (Scriber et al., 2005).
The newly described P. appalachiensis is most likely a more genetically diverged ‘mountain version’ of the ‘false second’ flights seen throughout the hybrid zone (Hagen & Lederhouse, 1985; Scriber, 2002a,b). Steep altitudinal gradients put P. appalachiensis in geographic parapatry with both parental species. However, its allochronic post-diapause emergence and its particular thermal zone affiliation appear to facilitate reproductive isolation from both parental species; the univoltine P. canadensis at the cooler side of the hybrid zone (2300–2600 °F degree-days), and the bivoltine P. glaucus at the warmer side of the hybrid zone (2600–2900 °F degree-days). Divergence between two species at neutral loci linked to traits under selection (possibly Ldh and Pgd in Papilio) will depend on their recombination distances and strength of selection (Charlesworth et al., 1998; Harrison, 1998; Wu, 2001). However, the allozymes themselves may be under local selection (Szymura & Barton, 1991; Wu, 2001). Different forms of Ldh enzymes have different kinetic properties influencing hatching times, developmental rates, performance, and mortality at different temperatures (Schulte et al., 1997), and a strong latitudinal cline for Ldh occurs at similar latitudes for fish Fundulus heteroclitus (Bernardi et al., 1993) as with our terrestrial Papilio. The newly detected rare allele (Ldh-20 hybrizyme) may have a unique functional role, or be closely linked to the gene regulating diapause induction or diapause termination in this hybrid swarm. Selection on Pgd-100 (directly or indirectly, via X-linkage) appears in areas with lower thermal accumulations (2250–2550 °F degree-days; Figure 5) than Ldh-100. However, it is still not clear whether these different selection pressures are driven by extrinsic (ecological) factors or intrinsic factors (e.g., genetic incompatibilities in recombinant genotypes such as Haldane effects on pupal mortality, especially in recombinant individuals with canadensis-type Pgd allozymes; Hagen & Scriber, 1995; see also Presgraves et al., 2003).
Ecological selection based on seasonal thermal unit constraints has apparently resulted in different thermal landscape niches for all three species of tiger swallowtail butterfly species (reflecting the different diagnostic Ldh and Pgd allele combinations in the eastern half of the USA; Table 2). However, it remains uncertain whether: (1) divergent selection on the X-chromosome traits results totally from ecological selection in the field due to close linkage of the Ldh-100 allele with the facultative gene (non-diapause, od–) that results in suicidal direct development (unsuccessful attempts at a second generation) anywhere with fewer than 2800 °F degree-days; or (2) or whether divergence is partially due to innate genetic incompatibilities in the hybrids (e.g., incompatible recombinant segments of X-chromosomes, or deleterious X–Y interactions, or unfit X-autosome or Y-autosome interactions; see Jiggins et al., 2001a,b). The post-zygotic egg and larval fitness of hybrid and backcross genotypes, while excellent for most reciprocal pairing combinations, may not extend to mating fitness. We have not yet field-evaluated the hybrid or backcross mating preferences, but experimental evidence regarding first or last male sperm precedence in these multiply-mating species has no clear pattern (Stump & Scriber, 2004).
This research was supported in part by the Michigan Agricultural Experiment Station (Project no. 01644) and the National Science Foundation (DEB-9201122 and DEB 9981608). Special thanks are extended to Dave Winter for his assistance with obtaining data from the Massachusetts Butterfly Atlas Project, and to William Houtz, Jim Maudsley, Howard Romack, and John Serrao for data and specimens from their collections. The scientists at Zedex, Inc. (Joe Russo, Jay Schlegel, and others) were especially helpful in aggregating the weather data from thousands of reporting stations into a GIS display format for total degree-days. Harry Pavulaan and David Wright generously provided the P. appalachiensis (and some P. glaucus) from West Virginia and Pennsylvania in 2003 and 2004. We thank Barb Gary, Holly Hereau, Matt Lehnert, Kent McFarland, Rodrigo Mercader, Michelle Oberlin, Laura Palombi, and Martha Smith Caldas for assistance in the laboratory and field. Special thanks are extended to the very helpful suggestions of an anonymous reviewer.