Cladistic analysis of 25 species of Dolichovespula and Vespula (yellowjackets) is used to investigate the evolution of social parasitism in these genera. Three species of yellowjackets are social parasites, or inquilines; that is, lacking a worker caste, and dependent on usurping the colony of a host species to obtain a worker force. Emery's Rule states that social parasites are more closely related to their hosts than to any other species. By investigating the phylogenetic relationships among the parasites and their hosts using cladistic analysis, we attempted to determine if Emery's Rule applies to yellowjackets, as is thought to be the case for the ants on which the theory was based. Sixty-eight morphological and behavioral characters are presented to resolve the phylogenetic relationships among these species. Cladistic analysis does not support Emery's Rule, because social parasites are not more closely related to their hosts than to any other species.
Introduction
Partly due to their eusocial behavior, vespine wasps of the genera Vespula and Dolichovespula, or yellowjackets, have been the focus of numerous behavioral studies. Several species are known to exhibit socially parasitic behavior, one of the most interesting offshoots of the evolution of social behavior. In social parasitism among insects, a female enters a colony and usurps the reproductive role of the resident queen, yet retains the worker force of the resident. In its extreme form, workers are not produced by the parasitic species. Such species, called inquilines, have no worker caste and do not build nests, rather they invade the colony of a host species and supplant the queen in order to obtain workers to rear their own brood. The peculiar inquiline behavior has sparked interest in various aspects of social parasitism, particularly in the mechanism by which such species have evolved (e.g., Emery, 1909; Wheeler, 1928; Taylor, 1939; de Beaumont, 1958; Buschinger, 1970, 1986, 1990; Wilson, 1971; Richards, 1971; West-Eberhard, 1986, 1996; Matsuura and Yamane, 1990; Hölldobler and Wilson, 1990; Bourke and Franks, 1991; Heinze, 1991).
Yellowjacket Inquilines
Three species3 of yellowjackets are inquilines: Vespula austriaca (hosts: V. rufa and V. acadica; Robson, 1898; Reed et al., 1979), Dolichovespula omissa (host: D. sylvestris; Weyrauch, 1937), and D. adulterina (hosts: D. arenaria, D. saxonica, and D. alpicola4; Wheeler and Taylor, 1921, de Beaumont, 1944; Wagner, 1978). Forty-three species are presently recognized in the genera Vespula and Dolichovespula (Carpenter and Kojima, 1997; Eck, 1998; Archer, 1999; Dong et al., 2002, 2005), and thus the inquiline behavior is uncommon (about 7% of the species). The inquiline species have long been recognized as taxonomically distinct. Schmiedeknecht (1881) placed V. austriaca in the new subgenus Pseudovespa because it lacks a worker caste, and Bischoff (1931) treated Pseudovespa as a genus. Bischoff diagnosed the genus on morphological grounds and added to it a new subgenus, Pseudovespula, to accommodate D. omissa and D. adulterina. He stated (Bischoff, 1931: 336–337) that each inquiline is most closely related to its host, as shown by (unspecified) similarity of male genitalia. Bequaert (1932) instead divided the yellowjackets into species groups and placed each inquiline in the same species group as its host. The concept of a polyphyletic genus for inquilines was followed by Blüthgen (1938), but he later (Blüthgen, 1943) broke up the taxon, recognizing two genera of vespine inquilines. Other taxonomists did not generally regard the diagnostic traits as sufficient to warrant high categorical rank for the inquilines, instead preferring to treat these taxa as subgenera (e.g., de Beaumont, 1944) or dispensing with separate categories altogether (e.g., Guiglia, 1948, 1972).
De Beaumont (1944, 1958) first depicted his views on relationships among the species of European vespines in the form of a branching diagram, although he did not specify the characters. Vespula austriaca was shown as most closely related to its host V. rufa, but D. adulterina was shown as less closely related to its host D. saxonica than the latter was to D. norwegica, and the position of D. omissa was doubtful with respect to its host D. sylvestris. Guiglia (1948, 1972) placed D. omissa in the same subgenus as D. adulterina, D. saxonica, and D. norwegica; the host D. sylvestris was placed in another subgenus. Later treatments of Vespula on a world basis left the relationship of V. austriaca and V. rufa unspecified (Yamane et al., 1980; Archer, 1981a, 1997), while Guiglia's arrangement has usually been followed in the species groups recognized in Dolichovespula (e.g., Archer, 1981b, 1999).
Evolution of Social Parasitism: Emery's Rule
Apparent close morphological resemblance between parasite and host led to the classical explanation of the origin of socially parasitic species, known as Emery's Rule. Emery's (1909) Rule states that parasites are more closely related to their host species than to any other species. Acceptance of Emery's Rule subsequently inspired the hypothesis that social parasites originate from their hosts by sympatric speciation (Buschinger, 1970, 1990; Bourke and Franks, 1991). However, the assumption that Emery's Rule is generally true has only recently been evaluated through cladistic tests (Ward; 1989, 1996; Agosti, 1994; Carpenter et al., 1993; Choudhary et al., 1994; Carpenter 1997; Danforth, 1999).
In Vespidae, the inquiline species in the paper wasp genus Polistes of the subfamily Polistinae have been shown to be a monophyletic group (Carpenter, 1997; see also Carpenter et al., 1993, and Choudhary et al., 1994). Within the subfamily Vespinae, cladistic treatment of supraspecific taxa (Carpenter, 1987b) has shown that the yellowjackets, the genera Vespula and Dolichovespula, are sistergroups, and several well-supported species groups can be recognized within these genera. But relationships among the species have not yet been comprehensively investigated, and the phylogenetic relationships among the inquilines and their hosts are thus unclear.
There has however been one previous noncladistic, “phylogenetic” analysis that discussed whether Emery's Rule applies to yellowjackets. Varvio-Aho et al. (1984) analyzed allozymes for eight species of European yellowjackets using distance methods, and concluded that the inquilines were most closely related to their hosts. Carpenter (1987a) showed that the analysis by these authors was flawed and that their data are in fact largely uninformative regarding the relationships of these yellowjacket species. Reanalysis of the data (Carpenter, 1987a) indicated that parasites in Dolichovespula were separated from their hosts. In this reanalysis, 50 trees of lower percent standard deviation than those that Varvio-Aho et al. found showed D. omissa more closely related to D. saxonica than to its host D. sylvestris.
Evolution of Social Parasitism: Usurpation in Yellowjackets
Another species of yellowjacket, V. squamosa, displays a different kind of social parasitism. It is evidently capable of founding its own colonies and has a worker caste, but in parts of its range it may usurp nests of other species. It ranges from New York to Guatemala, and in the eastern United States it has been reported to usurp V. vidua (Taylor, 1939), V. maculifrons (MacDonald and Matthews, 1975), and V. flavopilosa (MacDonald et al., 1980); additionally, Greene (1991) mentioned unpublished observations showing usurpation of V. germanica and Vespa crabro. Vespula squamosa has even been characterized as an obligate usurper of the nests of other species (Matthews, 1982), but this can evidently only be true in the northern part of its range (MacDonald and Matthews, 1984), because all of its host species are less widespread, and no other vespine species reaches as far south (Guatemala; see Carpenter and Kojima, 1997).
Usurpation of conspecific nests is common in yellowjacket queens, and may be a manifestation of competition for a nest site (Greene, 1991). Any brood, or even workers, present at the time of such usurpation may be “adopted” by the invading queen. Interspecific usurpation also occurs, but much less frequently. It has been reported in V. germanica (on V. vulgaris; Nixon, 1935), V. pensylvanica (on V. vulgaris; Akre et al., 1977), V. flavopilosa (on V. maculifrons and V. vulgaris; MacDonald et al., 1980), D. norwegica (on D. sylvestris; Edwards in Bunn, 1982: 172), and D. arenaria (on V. vulgaris; O'Rourke and Kurczewski, 1984).
Whether intra- or interspecific, such usurpation is considered a temporary form of social parasitism, in that the worker force produced by the usurping queen eventually replaces that of the host. Inquiline behavior may thus be considered an extreme form of usurpation, being permanent, with loss of the worker caste. This was the reasoning underlying the scenario for the evolution of social parasitism in wasps proposed by Taylor (1939). He suggested a progression through four stages: (1) intraspecific, facultative, temporary parasitism; (2) interspecific, facultative, temporary parasitism; (3) interspecific, obligate, temporary parasitism; (4) interspecific, obligate, permanent parasitism.
Taylor knew of the inquiline species and that intraspecific usurpation occurred, and he observed nest usurpation of V. vidua by V. squamosa. He considered that usurpation to be facultative, and thus three stages in his progression were observed in nature. Stage 3 was considered to be a theoretically necessary transition, with the usurping queen losing its ability to found a new colony but still retaining a worker caste. As mentioned before, this has been argued in fact to be the case for V. squamosa in part of its range.
Emery's Rule was a part of Taylor's arguments, as he stated (Taylor, 1939: 312), “Vespula squamosa and V. rufa var. vidua are closely related, both belonging to the subgroup of V. rufa (Bequaert, 1932). The two known social parasites are also closely related to their hosts.” The rufa group as construed by Bequaert (1932) comprised five species, a polytypic V. rufa consisting of no fewer than eight named forms. These forms later came to be treated as distinct species (Miller, 1958, 1961), and subsequent cladistic analysis (Carpenter, 1987b) showed that V. squamosa and V. sulphurea together formed the sister-group of the remaining species placed by Bequaert in the rufa group. The other hosts of V. squamosa, V. maculifrons and V. flavopilosa, are placed in the vulgaris species group (Bequaert, 1932; Jacobson et al., 1978), to which V. squamosa is even less closely related (Carpenter, 1987b). Although still considered to represent a key step in the evolution of social parasites, whether facultative or obligate, V. squamosa came to be characterized as an exception to Emery's Rule (e.g., MacDonald and Matthews, 1975; Edwards, 1980; Matsuura and Yamane, 1990; Greene, 1991).
Further elucidation of phylogenetic relationships among yellowjacket species is necessary to assess the applicability of Emery's Rule to yellowjackets. Cladistic analysis of morphological and behavioral characters is presented below, which attempts to resolve the relationships between parasitic and host species.
Materials and Methods
Table 1 presents a character matrix for the North American, European and Japanese species of Vespula and Dolichovespula; the character list is detailed below. Specimens examined are deposited in the American Museum of Natural History. Morphological characters examined were drawn primarily from the previous study of supraspecific groups by Carpenter (1987b) and the keys to species by Bequaert (1932), Guiglia (1972), Jacobson et al. (1978), Wagner (1978), Yamane et al. (1980), Akre et al. (1981) and Archer (1989). A number of characters establishing the monophyly of yellowjackets are retained in table 1. These characters do not pertain to relationships among species but are included to show that yellowjackets indeed constitute a monophyletic group, as demonstrated by Carpenter (1987b). Several prior studies had argued that Dolichovespula is more closely related to Vespa than to Vespula (Greene, 1979; Archer, 1987), as did one subsequent paper (Schmitz and Moritz, 1990; see Carpenter, 1992). Morphometric differences were excluded: Archer's (1989) key used a few characters such as ocelloccipital length and relative length of the malar space among species of Dolichovespula, but we did not find the states to be well delimited (note that the inquilines differ from their host species in a number of such features; see Bischoff, 1931; Weyrauch, 1937; Eck, 1979; Reed and Akre, 1982). The matrix includes nine color characters, which are among the features most commonly used in keys to distinguish species. A few new characters from male genitalia are also included. Male genitalia were dissected, cleared slightly in lactophenol, and examined in glycerin. Morphological terminology is as in Carpenter (1987b). Behavioral characters were also taken from Carpenter (1987b), with specific scores taken from the reviews by Spradbery (1973), Edwards (1980), Akre et al. (1981), Akre and MacDonald (1986), Matsuura and Yamane (1990), Greene (1991), and according to the citations for certain characters detailed below. The outgroup is the hornet Vespa crabro, which is not closely related to yellowjackets (Carpenter, 1987b). Multistate characters were treated as additive where similarity among character states was observed to be nested; otherwise, such characters were treated as nonadditive. Cladistic analysis (Hennig, 1966) was implemented with the program NONA (Goloboff, 1999a). Character optimization and diagnoses were accomplished with the WINCLADA program (Nixon, 2002). Illustrations are provided for selected characters.
Table 1
Characters for Species of Yellowjackets and Hornet Outgroupa
Character List
Worker caste: present = 0; absent = 1.
Base of forewing 1RS cell: M oblique with respect to m-cu1 at base of second submarginal (1RS) cell = 0; M vertical (apex of discal cell truncate) = 1.
Placement of forewing m-cu2: far from r-m2 = 0; close to r-m2 = 1. This was character 4 in Carpenter (1987b), where it was expressed as the second submarginal cell with section of M vein distal to m-cu2 crossvein shorter vs. longer than section of RS vein basal to r2 crossvein. The sections of M along the posterior edge of the second submarginal cell are subequal in the derived state, while the distal section is much longer than the basal section in the plesiomorphic state.
Hamuli placement: basad of hindwing R1 and RS = 0; at fork of R1 and RS = 1.
Apical angles of clypeus: bluntly produced = 0 (fig. 1); sharply produced = 1 (fig. 2). This is usually cited as a key character of females in the inquline species, but the males also have the apical angles more sharply produced than males of the other species.
Clypeal punctation: close ventrally (interspaces less than diameter of punctures) = 0 (fig. 3); sparse (interpspaces greater than diameter of punctures) = 1 (fig. 4).
Occipital carina: running to base of mandible = 0 (fig. 5); effaced near mandible = 1 (figs. 6, 7); obliterated dorsally = 2 (fig. 8).
Malar space: short (length less than that of last flagellomere) = 0 (fig. 9); long (length greater than or equal to that of last flagellomere) = 1 (fig. 10).
Tyloides: two per flagellomere = 0; one, only on apical flagellomeres = 1; absent = 2.
Female mandibular flange: absent = 0 (fig. 11); present basally on outer posterior edge of mandible = 1 (fig. 5).
Margin behind the third mandibular tooth: distinctly concave = 0 (fig. 12); straight or at most slightly concave = 1 (figs. 13, 14).
Labial palpus: third segment with a strong seta = 0; without this seta, but with hairs = 1.
Pronotal carina: present = 0; dorsally reduced = 1 (figs. 15, 16); laterally effaced = 2 (figs. 17, 18).
Pretegular carina: present = 0; absent = 1.
Pronotal striae: absent = 0 (fig. 15); present = 1 (fig. 16).
Anterolateral scutal depression: not developed = 0; present = 1 (fig. 17); with posterolateral extension = 2 (fig. 18). Although Vespinae do not possess notauli, these depressions follow the course of notauli in other vespids that possess them (Duncan, 1939).
Scutal lamella: rim of scutum raised into lamella beside tegula = 0; absent = 1.
Hindcoxal carina: absent = 0; present = 1.
Hindtibia: with long erect hairs = 0; without long erect hairs = 1.
Dorsum of metasomal tergum I: flat in lateral view = 0; with slight depression behind anterior edge = 1.
Female metasomal sternum VI process: without dorsolateral process = 0; with dorsolateral process = 1.
Queen sting: straight = 0; recurved = 1. It has long been known that inquiline females have the sting recurved in comparison to their hosts (Weyrauch, 1937). Reed and Akre (1982: fig. 2E) noted a sharp distal bend in the sting of V. squamosa; however, this is evidently not found in all queens of this species.
Ovariole number: 7 = 0; 6 = 1. See Kugler et al. (1976).
Male metasomal tergum VII: disc evenly convex in profile = 0; disc depressed = 1; disc depressed steplike = 2.
Male metasomal tergum VII apex: rounded = 0; emarginate = 1.
Male metasomal sternum VII apex: rounded = 0; transverse = 1; emarginate = 2 [nonadditive].
Apex of aedeagus: transverse, projecting laterally = 0; not projecting laterally = 1; subcircular = 2.
Subapical expansion of aedeagus: absent = 0; present, about as broad as shaft of aedeagus = 1; present, broader than shaft of aedeagus = 2.
Thickened margins of aedeagus: not projecting apicad = 0; projecting apicad = 1.
Aedeagal apical indentation: deep = 0; shallow = 1; absent = 2.
Width of aedeagus: broad (aedeagus as wide or wider apically as medially) = 0; narrower apically than medially = 1; attenuate, much narrower apically than basally = 2.
Aedeagal apical lobes: absent = 0; present = 1.
Aedeagal medial lobes: absent = 0; present = 1.
Paramere process: paramere with broad inflection near spine = 0; broad pointed process = 1; process distally prolonged = 2; narrowed distally = 3; absent = 4 [nonadditive].
Paramere base: not emarginate dorsally = 0; emarginate dorsally = 1.
Volsella: long (extending at least to apex of parameral spine and aedeagus) = 0; short (not extending as far as apex of parameral spine or aedeagus) = 1.
Volsella dorsal lobe: absent = 0; present = 1.
Digitus: large, pincer-shaped lobe = 0; very slender, fingerlike lobe = 1.
Hairs on parameral spine and digitus: present dorsally = 0; absent dorsally = 1.
Pale markings: yellow = 0; white = 1.
Yellow eye loop: absent = 0; present = 1.
Genal band: continuous = 0; interrupted = 1.
Scape ventrally: pale = 0; dark = 1.
Two longitudinal yellow stripes on scutum: absent = 0; present = 1.
Lateral surface of the thorax: without yellow hairs = 0; with yellow hairs = 1.
Metasomal tergum I hairs: pale = 0; dark = 1.
Metasomal terga I and II maculations: present = 0; terga entirely black = 1.
Reddish markings on metasomal tergum I or II: absent = 0; present = 1.
Larval mandible: tridentate = 0; single apical tooth, ill-defined dorsal tooth = 1.
Larval clypeus: frontoclypeal suture present = 0; absent = 1.
Larval spiracle: collar processes simple = 0; branched = 1.
Nest: nesting site aerial = 0; nesting site in cavity or underground = 1. See Akre et al. (1981, 1982) on variation in nesting site.
Paper type: gray, pliable = 0; tan, fragile = 1. See Greene (1979) on variation in D. maculata.
Scalloping: absent, envelope laminar = 0; elongate = 1; imbricate = 2 [nonadditive]. See Greene (1979).
Suspensoria: comb suspensoria pillarlike = 0; ribbonlike supporting first comb = 1; interconnected and ribbonlike throughout = 2 [nonadditive].
Worker-cell comb: multiple combs of worker cells = 0; one comb of worker cells = 1.
Mixed cell combs: present = 0; worker and queen cells on different combs = 1. The scoring for V. flaviceps and V. shidai is based on Kojima (personal commun.).
Attachment of embryo nest: pedicel of embryo nest with simple cylindrical shape = 0; thin and twisted, with hanging sheet = 1. See Yamane and Makino (1977), Yamane et al. (1981) and Matthews et al. (1982) on architecture of embryo nests. Yamane and Makino (1977: 26) and Yamane et al. (1981: 496, footnote) allude to some confusion in the literature about the state in V. vulgaris, but Matthews et al. (1982) do not describe it as different from other species of Vespula.
Pedicel of embryo nest: coated with glossy oral secretion = 0; uncoated = 1.
Embryo nest disc: absent = 0; present = 1.
First sheet of embryo nest: bonded to substrate = 0; bonded to hanging sheet = 1; bonded to pedicel = 2 [nonadditive].
Expansion of envelope of embryo nest: later sheets beginning on previous sheet = 0; later sheets independent = 1. Yamane et al. (1981) considered this character to be an important difference between Vespula and Dolichovespula, although it has not been studied for many species. In addition to the species they listed, figures in Duncan (1939) show the states in a few species. On the other hand, the figures in Brian and Brian (1948), cited by Yamane et al. (1981) as showing later sheets beginning on the previous sheet in D. sylvestris, actually appear to show variation in this character (cf. Brian and Brian, 1948: figs. A and B).
Vestibule: absent = 0; present = 1.
Colony size: “small” (<2500 cells and 400 workers at peak) = 0; “large” (3500 cells and 500 workers or more) = 1. Ranges overlap in many species (see Akre et al., 1981).
Hibernaculum: in pre-existing cavities = 0; partly dug by queen = 1. See Matsuura and Yamane (1990).
Royal court: “royal court” of workers present, surrounding queen = 0; no distinct royal court of workers surrounding queen = 1. See Matsuura and Yamane (1990).
Prey: live arthropods = 0; also vertebrate carrion = 1. This behavior varies more than indicated in the matrix; see Akre et al. (1981).
Larval hunger signal: consistent, audible cell wall scraping = 0; low frequency scraping = 1; inaudible = 2 [nonadditive]. See Matsuura and Yamane (1990).
Figs. 1–6
Characters of yellowjackets. 1–2, clypeus. 1, V. rufa (Linnaeus). 2, Vespula austriaca (Panzer). 3–4, head in frontal view. 3, Dolichovespula sylvestris (Scopoli). 4, D. norvegicoides (Sladen). 5–6, head in lateral view. 5, V. pensylvanica (de Saussure). 6, V. vidua (de Saussure). All scale bars equal 1 mm.
Figs 7–12
Characters of yellowjackets. 7–8, head in lateral view. 7, Dolichovespula norvegicoides (Sladen). 8, D. adulterina (du Buysson). 9–10, head in frontal view. 9, Vespula rufa (Linnaeus). 10, D. alpicola Eck. 11, head in lateral view, D. arenaria (Fabricius). 12, mandibular teeth, V. germanica (Fabricius). All scale bars equal 1 mm.
Figs 13–18
Characters of yellowjackets. 13–14, mandibular teeth. 13, Vespula shidai Ishikawa, Yamane and Wagner. 14, Dolichovespula saxonica (Fabricius). 15–18, mesosoma in lateral view. 15, D. adulterina (du Buysson). 16, Dolichovespula maculata (Linnaeus). 17, V. rufa (Linnaeus). 18, V. germanica (Fabricius). All scale bars equal 1 mm.
Results
Analysis of the combined morphological and behavioral characters using the mult* (random addition of terminals during tree construction, with up to 100 replications, each of which hit minimum length) and max* (branch swapping by tree-bisection-reconnection) commands of NONA resulted in one cladogram (fig. 19; 106 steps, consistency index = 0.79, retention index = 0.93), which is stable to either successive weighting using the rescaled consistency index and NONA or implied weighting as implemented in the program PIWE (Goloboff, 1999b).
Fig. 19.
Cladogram resulting from analysis of all the characters in Table 1. Characters have been optimized with only unambiguous changes plotted. Character numbers are above the hashmarks; state changes are shown below, with the respective primitive and derived conditions separated by a “>”. Filled hashmarks denote uncontroverted changes, whereas open hashmarks indicate homoplasy in the character. Characters supporting the monophyly of yellowjackets as a whole are not plotted. This cladogram also results from analysis of only the morphological characters in Table 1.
The morphological and behavioral characters are not significantly incongruent, according to the incongruence length difference test (Farris et al., 1994) as implemented in WINCLADA (1000 replicates, p = 0.4016 with autapomorphies not deleted and the default “amb- poly = ” setting; p = 0.3876 with “amb = poly-”). Analysis of the 51 morphological characters resulted in the same cladogram as in figure 19 (81 steps, consistency index = 0.77, retention index = 0.93), which is stable to either successive or implied weighting. By contrast, analysis of the 17 behavioral characters produced 1705 cladograms (24 steps, consistency index = 0.87, retention index = 0.95), only 343 of which are strictly supported (see Nixon and Carpenter, 1996), but the consensus of either set of cladograms is entirely unresolved. This is due to the fact that the inquiline species are scored as missing or inapplicable for all the behavioral characters. When those three species are deleted and the behavioral characters reanalyzed, one cladogram resulted (fig. 20, 24 steps), differing from the topology supported by either the combined or morphological characters primarily in being less resolved. Dolichovespula and Vespula are both supported, but the former is not otherwise resolved, while within Vespula the rufa and vulgaris groups are supported, but V. squamosa and V. sulphurea are grouped with the vulgaris group, not the rufa group. This is the only point in which the morphological and behavioral characters disagree, and the combination of characters gives the result supported by morphology. The placement of V. squamosa and V. sulphurea in the behavioral result is supported only by character 66 (collection of carrion in addition to live prey), and, as noted above, reports of this behavior vary more than indicated in the matrix. In general the behavioral characters suffer more ambiguity, due primarily to lack of observations: the behavioral characters comprise 442 cells in the matrix of table 1, 24.2% of which are scored with missing values, 10.4% of which are inapplicable, and 5.88% of which are polymorphic. Thus, “real” values are observed in only 59.5% of the cells. The morphological characters have only 2.63% missing values and 2.11% polymorphisms, and thus 95.24% of the 1326 cells are “real” values.
Fig. 20
Cladogram resulting from analysis of the behavioral characters in Table 1, with the three inquiline species deleted.
The combined characters do not completely resolve the interrelationships among the species (fig. 19). According to the cladogram, however, the social parasites of both Dolichovespula and Vespula have evolved separately from their hosts: D. omissa and D. adulterina are more closely related to each other than they are to their hosts, D. sylvestris for D. omissa, and D. arenaria, D. saxonica and D. alpicola for D. adulterina. Vespula austriaca is the sister-group to a clade comprising the remaining species of the V. rufa species group, which includes the host species V. rufa and V. acadica. Even a looser formulation of Emery's Rule, to allow for subsequent speciation (Ward, 1996), in which a clade of social parasites has as nearest non-parasitic outgroup a clade that includes the host species, is not applicable to Dolichovespula. Dolichovespula sylvestris is the sister-group to D. omissa (its parasite) + D. adulterina, but the hosts of the latter are scattered in a paraphyletic grade of other Dolichovespula species.
Of course, an even looser formulation (Danforth, 1999), in which a social parasite clade renders the host group paraphyletic, does apply to the yellowjacket inquilines—but such a statement has very little empirical content, because it could apply to a myriad of possible paraphyletic “groups”.
It is worth commenting that the monophyly of D. adulterina + D. omissa is established by four synapomorphies, three of which (i.e., character 1, absence of a worker caste;, character 5, produced apical angles of the clypeus;, and character 22, recurved sting) are also found in the other inquiline species, V. austriaca. These features might thus be considered “morphological accoutrements of parasitism” (West-Eberhard, 1996: 316), and thereby be dismissed as convergent adaptations. Whereas such an argument may be advanced for characters 1 (which is an expression of inquiline behavior) and 22 (which is used in combat during usurpation), the argument is less defensible for character 5, because the apical angles of the clypeus are produced in males, less so than in females but more than in nonparasitic species. A character expressed in males seems less obvious as a candidate for an adaptation related to host domination, unlike features not included in the matrix of table 1 such as enlarged mandibles and femora (see Weyrauch, 1937; Reed and Akre, 1982). But that aside, the fourth synapomorphy for D. adulterina + D. omissa (i.e., character 7, occipital carina effaced dorsally) does not appear to be at all related to parasitism—and is not found in V. austriaca. The monophyly of D. adulterina + D. omissa thus is established not just by characters that are possible parasitic convergences—and moreover it is not rejected by the other characters analyzed here.
Discussion
As shown by this study, social parasites are not more closely related to their host than to any other species, and hence Emery's Rule is not applicable in yellowjackets. Other studies have rejected the applicability of Emery's Rule to paper wasps (see Carpenter, 1997). The utility of Emery's Rule in wasps is thus diminished altogether.
To the extent that this and other studies fail to support Emery's Rule in wasps, the theory that social parasites evolve sympatrically from their hosts is likewise undermined. But although Emery's Rule is not supported in wasps, nor in ants (Agosti, 1994; Ward, 1989, 1996) or bees (Williams, 1994; Danforth, 1999) where it has been tested, it has not been tested in most ant and bee social parasites. Social parasitism has arisen numerous times in ants, bees and wasps, and so sympatric speciation cannot be rejected out of hand. However, it should be clear that the basic tenets of the principle that have become generally known as Emery's Rule need to be reanalyzed and assessed carefully. Any rule that generalizes the way species evolve without regard to specific conditions is bound to be suspect and clearly needs to be tested before any further assumptions are made. If it seems plausible that in some cases the parasites evolve sympatrically from their hosts, in other cases the parasites just as plausibly evolved allopatrically, for example in relation to climatic factors (Richards, 1971; Matsuura and Yamane, 1990).
What then of Taylor's scenario for the evolution of inquiline behavior as a progression from conspecific queen competition? The extension of such usurpation into inquilinism may still seem a compelling explanation, despite the inapplicability of Emery's Rule (Carpenter et al., 1993). However, there is no clear phylogenetic progression. Grandcolas et al. (1994) have outlined how correspondences between types of phylogenetic patterns and theoretical evolutionary processes may function as tests of the theories. For evolutionary scenarios, the “time lag” or order of appearance of features may support or reject a suggested evolutionary progression based on the association of those features. This approach was applied to the origin of inquiline behavior in Polistes by Carpenter (1997), by cladistic optimization of behavioral traits on a cladogram for the inquiline species and their hosts. The optimizations accorded with part of Taylor's scenario, specifically that (obligate and permanent) interspecific usurpation evolved from (facultative and temporary) intraspecific usurpation. Stage 1 in Taylor's scenario thus preceded stage 4. But other parts of the scenario were not supported; specifically, it is unclear whether (facultative and temporary) interspecific usurpation preceded inquiline behavior. Inquiline behavior could have evolved directly from intraspecific usurpation (stage 1) or via an intervening stage of temporary interspecific usurpation (stage 2). Furthermore, the third stage of the scenario is not observed in Polistes.
A similar optimization of usurpation behavior in yellowjackets is shown in figure 21. State 0 is intraspecific usurpation; state 1 is facultative, temporary usurpation; state 2 is obligate, temporary usurpation (for which V. squamosa is scored as polymorphic); and state 3 is obligate, permanent usurpation (inquiline behavior). The variable is treated as additive. Species with a question mark are those for which we have not been able to find reports of usurpation. The optimization that is shown plots only unambiguous changes, with the state preceding the “<” symbol being the state inferred by the optimization for taxa with missing values. As with Polistes, the behavior of the inquilines may have evolved directly from intraspecific usurpation (possible with D. adulterina + D. omissa, where the optimization is ambiguous) or from facultative interspecific usurpation (V. austriaca). Obligate temporary usurpation evolved from facultative interspecific usurpation only within V. squamosa (not plotted). It is not part of a progression to inquiline behavior, and the conclusion remains the same even if V. squamosa is scored as monomorphic for state 2. Not all of the stages of Taylor's scenario may thus be necessary in the evolution of social parasitism. The tactics used by the inquilines during usurpation are known to differ among the species: D. adulterina is passive when invading the nest of D. arenaria, assuming a submissive posture during attacks and coexisting with the D. arenaria queen for a time (Evans, 1975; Jeanne, 1977; Greene et al., 1978), while V. austriaca immediately attempts to kill the V. acadica queen of an invaded nest (Reed and Akre, 1983). The usual timing of invasion is also different, whether before emergence of the first workers (D. adulterina) or after (V. austriaca). Inquiline behavior, which originated more than once in yellowjackets, has thus undergone more than one evolutionary pathway. A single, simple scenario may therefore not be universally applicable.
Acknowledgments
This study was begun while the second author was at Columbia College, Columbia University, working under the auspices of the first author at the American Museum of Natural History. EPP thanks W. Bock, D. Agosti, S.J. Gould, R. Dawkins, R. Holloway, M. Stiassny, J. Glover, E.S. Scott, J. Cole, C.L. Cosby, J.W. Wenzel, Z. Liu, M.E. Archer, J. Kojima, and P. Gambino for moral support, lively philosophical debate, spiritual guidance, and finally for reading and commenting on the manuscript. Use of AMNH facilities was partly supported by NSF Award 9424068 to Melanie Stiassny.
References
Notes
[1] The species taxonomy of Vespinae is historically quite complicated, with most recent literature recognizing a fourth inquiline species, the Nearctic D. arctica. We follow here the taxonomy used by Carpenter and Kojima (1997), which treats D. arctica as a synonym of the Palearctic D. adulterina.
[2] D. norwegica is cited as a host as well in some reviews (e.g., Akre, 1982; Matsuura and Yamane, 1990), but this is based partly on treatment of D. saxonica as a synonym of D. norwegica. The positive host records in Europe are limited to D. saxonica, with indirect evidence suggesting D. norwegica (Guiglia, 1972: 136).
