Biochemical Studies on the Relationship Between Socially Parasitic Ants and Their Hosts

JORGEN HEINZE Zool. Inst. II, R6ntgenring 10, 8700 WQrzburg, F.R.G.

Key Word Index--Doronomyrmex; Harpagoxenus; Leptothorax; Formicidae; Hymenoptera; social parasites; Emery's rule; electrophoresis. Abstract--Analysis of enzyme patterns suggests close phylogenetic relationships between socially parasitic ants of the genera Harpagoxenus and Doronomyrmex and their Leptothorax hosts. Doronomyrmex goesswa/di and D. kutteri are indistinguishable in enzyme patterns from their host, Leptothorax acervorum. A newly discovered workerless parasite from Canada, L. paraxenus, however, appears to be more closely related to other non-parasitic species than to its host, L. sp.B (= L. canadensis?). The results are discussed with respect to current hypotheses on the evolution of social parasites.

Introduction Ants are among the most successful and most abundant insects on Earth. They can be found in the most diverse habitats, and they build their nests almost everywhere: in soil, litter, tree stumps, rock crevices, the mounds of termites and the colonies of other ant species. Among the latter are the social parasites: ants which are no longer capable of feeding themselves or tending their own larvae, but which instead rely on the help of workers from other species. About three per cent of the approximately 9000 ant species are known to lead a permanent or temporary parasitic life [1]. The freshly inseminated queens of social parasites invade the colonies of other species which they parasitize by begging food and letting the hosts take care of their brood. Queens of the north and central European slave-maker Harpagoxenus sub/aevis, for example, kill or expel the host queen and the adult workers. Workers which eclose from the conquered host brood forage, provide the parasites with food, and nurse the Harpagoxenus larvae. The original host workers die within a year or two and have to be replaced. Hence, Harpagoxenus workers perform slave-raids, during which they pillage neighbouring colonies of the host species and carry away their pupae. The queens of some workerless parasites, such as Doronornyrmex kutter/and D. goesswa/dl; sneak into a host nest, where they are tolerated by the resident ants. They produce sexual offspring but none of their own workers. Whereas D. kutter/is an inquiline, which peacefully lives alongside the fertile host queens, D. goesswa/d/'is a host queen-intolerant parasite. For more than a hundred years biologists have speculated on the evolutionary origin of social parasitism. With the exception of guest ants (see below), social parasites are often morphologically very similar to their hosts, and in 1909 Carlo Emery concluded that "the slave-making, temporarily and permanently parasitic ants orig- inate from closely related forms which serve them as hosts" [2]. Today, a loose version of "Emery's rule'--parasitic ants and their hosts are closely related--is widely accepted. It has also been suggested that in particular the workerless parasites have evolved intraspecifically from their host species [3-5]. Permanent social parasites are extremely rare. Several species have been found only once and, despite extensive and repeated searches at the typical localities, only museum material is available. The close relationship of parasites and hosts therefore has usually been inferred from

(Received 16 November 1990)

195 196 J. HEINZE

morphological resemblance rather than detailed phylogenetic analysis [6]. Though electrophoretic data are now commonly used to reconstruct phylogenies of social insects [7], genetic investigations of parasites and their hosts are still comparatively rare. In wasps and bees, it has been inferred from enzyme polymorphisms that cuckoo bumble bees (Psithyrus spp.) are a monophyletic group which radiated secondarily to several host species [8], and that the parasitic wasps are of polyphyletic origin [9] (but see [10, 11]). In ants, the distribution of esterase allozymes suggests that small queens (microgynes) in the nests of Myrmica rubra are genetically isolated from large queens (macrogynes) and presumably belong to a separate, parasitic species [12]. During the last few decades, several teams have intensively studied the parasite- rich myrmicine tribe Leptothoracini. In suitable habitats, these little and inconspicuous ants form dense populations and, in some places, parasites can regularly be collected. Thus, Leptothorax parasites are among the best studied socially parasitic ants, and they are the ideal species for a first biochemical approach toward Emery's rule. Electro- phoresis has already been used in a number of leptothoracine species to estimate the relatedness between nestmates [13-15] and to clarify the taxonomic position of species [16, 17]. Inter- and intraspecific genetic diversity is low, and esterase or dehydrogenase isozymes have successfully been used to characterize taxa or to distin- guish between sibling species [17-19]. As part of a study on the systematics of the nearctic Leptothorax "muscorum" complex, it was possible to collect data on the variability of enzymes in a number of socially parasitic species and their hosts. A first qualitative analysis of esterase banding patterns [19] had already supported recent ideas for a systematic revision of the whole tribe [20, 21]. In this paper, more detailed electrophoretic data from parasitic Leptothoracini and their hosts, especially of the genera Leptothorax (s.str.), Doronornyrmex, and Harpagoxenus, are presented. In temperate North America and Eurasia, the complex genus Leptothorax is represented mainly by three subgenera: Leptothorax(s.str.) (= Mychothorax), Myrafant, and Temnothorax.Most of 350 or more named taxa of Leptothoraxare ordinary non- parasitic species. A few are workerless parasites and one species, L. duloticus, is a slave-maker (Table 1). The genus Doronomyrmexcomprises only workerless parasites of L. (s.str.) (with the exception of the dubious D. pocahontas), and the three species of Harpagoxenus available in this study are active slave-makers, with H. canadensisand H. sublaevis enslaving colonies of Leptothorax(s.str.), but H. americanusfound only in nests of Myrafant. Finally, Formicoxenus quebecensis is a guest ant and lives in the nests of Myrrnica alaskensis. Guest ants can be found in colonies of various ant species, to whom they are not closely related. They beg food from their hosts but take care of their brood themselves. The genus Formicoxenusbelongs to the Leptothoracini and has been thought to be closely attached to the subgenus Leptothorax(s.str.) [22].

Materials and Methods Complete colonies of host species were collected during the past five years in numerous localities in North America and Central Europe. Social parasites were collected in the following sites: Doronomyrmex kutteri: Nyehusen, Sweden; Doronomyrmex goesswaldi and D. pacis: La Villette, France; Doronomyrrnex pocahontas: Maligne Canyon (Jasper N. P., Alberta); Harpagoxenus canadensis: St. Sim6on (Comt~ de Charlevoix-Est, Quebec), Tadoussac (Saguenay Co., Qua.), Rouyn-Noranda (Temiscamingue Co., Qua.), MacKenzie Mountain (Inverness Co., Nova Scotia); /4. sublaevis: Nfirnberger Reichswald, F.R.G.; Leptothorax paraxenus; Milton, Ontario; L. wilsoni: Moncton (Westmoreland Co., New Brunswick); Formicoxenus quebecensis: Waswanipi (Co. de Abitibi, Qua.). Single workers, males and females were crushed individually in 40 ill running buffer with 15% glycerol and 0.01% bromothymol blue. Proteins were separated in 12.5-cm-long 7.5% polyacrylamide gels (gel buffers: 0.47 M Tris-HCI, pH 8.8 and 0325 M Tris-HCI, pH 8.0 for phosphoglucose isomerase (PGI); running buffer 036 M glycine, 0.025 M Tris, pH 8.3) at 10°C with a current of approximately 10-20 mA for 3 h, and on 7.5- or 13-cm-long cellulose acetate plates (Cellogel, Milano, and Helena Laboratories, Beaumont, Texas; gel buffer and running buffer: 0.01 M sodium phosphate/citrate, pH 6.4) with a constant voltage of 200 or 350 V for 1.5 h. Dehydrogenases were stained using the following reagents: 2 mg BIOCHEMICAL STUDIES ON PARASITIC ANTS AND HOSTS 197

TABLE 1. RANGE OF THE NON-PARASITIC SPECIES OF LEPTOTHORAX (S.STR.) AND OF THE SOCIALLY PARASITIC SPECIES OF LEPTOTHORAX, HARPAGOXENUS AND DORONOMYRMEX, BASED ON [1 ] AND AUTHOR'S OBSERVATIONS. SPECIMENS OF THE SPECIES PRINTED IN BOLD LETTERS WERE EXAMINED IN THIS STUDY

Non-parasitic Typical habitat Known range

L. acervotum Alpine and boreal Holarctic (Fabricius, 1973) coniferous forests occasionally in deciduous forests L. muscorum Alpine and boreal Palaearctic (Nylander, 1846) coniferous forests L. gredleri Light pine stands, rose and Central Europe (Mayr, 1855) blackthorn thickets L. scarnni Alpine coniferous forests Caucasus, Pontus (Ruzsky, 1905) L. craesipilis Pine and cottonwood forests U.S. Rocky Mts (Wheeler, 1917) L tetractus Alpine and boreal Nearctic (Francoeur, 1986} coniferous forests L spagnico/us Spruce bogs Central Quebec (Francoeur, 1986) L. sp.A Open alpine and boreal Eastern North America coniferous forests L. sp.B Boreal coniferous forests, Eastern North America occasionally in deciduous forests L. sp.C. Alpine coniferous forests Canadian Rocky Mts L. sp.D Alpine coniferous forests Canadian Rocky Mrs

Parasite Host species Known range Type of parasitism

L. faberi L. sp.D Jasper N. P., AIta. Inquiline (Buschinger, 1982) (type locality) L. wilsoni L sp.B. Quebec, New Brunswick, Workerless, queen-intolerant? (Heinze, 1989) New Hampshire L. pataxenus (Heinze and Alloway, in L sp.B Southern Ontario and Workerless, queen-intolerant? prep.) Quebec D. geesswa/di L. acervorurn Alps, S. Sweden Workerless, queen-intolerant (Kutter, 1967) D. kutteri L. acervorum Alps, S. Sweden, Inquiline (Buschinger, 1966) Central Germany D. pacis L. acervorum Alps Inquiline (Kutter, 1950) D. pocahontas L. sp.C Maligne Canyon, Alta. ? (Buschinger, 1979) (type locality) N. cenaden$is L. spp.A, B Eastern North America Slave-maker (M. R. Smith, 1939) H. subleevis L. acervorum, Central and Northern Slave-maker (Nylander, 1849) L. muscorum, Europe L gred/eri

Parasites of species belonging to the subgenus Leptothorax (MyrafantJ

H. arnericanus L. ambiguus, Eastern North America Slave-maker (Emery, 1895) L. curvispinosus, L. Iongispinosus L. duloticus L. ambiguus, Eastern United States Slave-maker (Wesson, 1937) L. curvispinosus, L. Iongispinosus L. minutissimu$ L. curvispinosus Eastern United States Inquiline (M. R. Smith, 1942)

NAD or NADP, 2 mg NBT, 0.2 mg PMS, and approx. 10 rng of the substrate dissolved in 5 ml of staining buffer (0.2 M Tris-HCI, pH 7.0, for lactate dehydrogenase (LDH), 0.5 M Tris-HCI, pH 8.0, for the other dehydrogenases). 198 J. HEINZE

Results and Discussion Genera~patterns of enzyme variabi/ity Though individual Leptothoraxants weighed less than 1 mg, it was possible to stain up to six or more different enzyme systems in a single, crudely homogenized adult. The following enzymes gave reproducible, though often weak stains in females, males and workers: ADH (alcohol dehydrogenase, EC 1.1.1.1), 0¢-GPDH (0¢-glycerophosphate dehydrogenase, EC 1.1.1.8), LDH (lactate dehydrogenase, EC 1.1.1.27), ~-HBDH (~- hydroxybutyrate dehydrogenase, EC 1.1.1.30), MDH-1 and MDH-2 (malate dehydrogenase, EC 1.1.1.37), ME (malic enzyme, EC 1.1.1.40), IDH (isocitrate dehydrogenase, EC 1.1.1.42), PGD (6-phosphogluconate dehydrogenase, EC 1.1.1.44), G6PDH (gtucose-6- phosphate dehydrogenase, EC 1.1.1.49), AO (aldehyde oxidase, EC 1.2.1.5), XDH (xanthine dehydrogenase EC 1.2.3.2), TO (tetrazolium oxidase or superoxide dis- mutase, EC 1.15.1.1), and PGI (phosphoglucose isomerase, EC 5.3.1.9). Some other enzymes, which could be easily stained in a Drosophi/a standard, gave no results in leptothoracine ants (e.g. glutamate-oxalacetate transaminase, EC 2.6.1.1, hexokinase, EC 2.7.1.1, alkaline phosphatase, EC 3.1.3.1), or were clearly visible only in larvae (leucine aminopeptidase, EC 3.4.11.1) or pupae (unspecific esterases, EC 3.1.1.1, especially Ester- ase-7 [19]). IDH and PGD could not be stained in polyacrylamide gels with a gel buffer of pH 8.0 or 8.8, but on cellulose acetate gels and ultrathin polyacrylamide gels for IEF. As has also been reported for esterases [19], enzyme patterns of Leptothorax (s.str.) and its parasites are rather uniform compared with those of Myrafant. Whereas in 26 Myrafant and Temnothorax from North Africa and Europe, 12 electromorphs of IDH have been separated in starch gels [16], in the present study only two tDH banding patterns could be found in 20 taxa of Leptothorax(s.str.) and its parasites (Table 2). The apparent difference in genetic variability might be caused in part by the use of different separation techniques, but ultrathin layer IEF showed similar results: only two IDH electromorph patterns in Leptothorax (s.str.), Harpagoxenus sub/aevis, H. canadensis, and Doronomyrmex, but at least five different patterns in eight Myrafant species. Similar results were obtained with other enzymes, e.g. TO, with four different electromorphs in eight Myrafantand only one in all Leptothorax (s.str.) (data not presented). In Leptothorax (s.str.) and its parasites, XDH, ADH, LDH, AO, TO, G6PDH appear to be uniformly monomorphic throughout the species. MDH-2 and ME were found to be polymorphic in the non-parasitic species, and they are also variable at least in Harpagoxenus sublaevis [15]. Of MDH-1, IDH and PGD, two or more electromorphs have been separated, and the different taxa were more or less fixed for one of them each [17]. Thus, specimens of L. acervorum from Sweden, Germany, France, Japan and Alaska all have a fast migrating MDH-1 with an approximate isoelectric point of 7.2, but more than 95% of all L. sp.B ( -- L. canadensis?) from New England and Eastern Canada have a stow electromorph (pl = 6.6) (Table 2). Similarly, the European Lepto- thorax (s.str.) and the nearctic species L. crass/pi/is and L. sp.A all have a fast migrating IDH, whereas L. sp.B. and L. retractus exclusively have a slow electromorph. Addition- ally, very rare allozymes have been detected only when large numbers of colonies from a single site are examined. Thus, in 80 colonies of L. muscorum from the Reichswald population in Southern Germany, workers from two colonies had hetero- zygote, three-banded MDH-patterns with a previously unknown, very fast migrating allozyme, and in one of 30 colonies of L. sp.B from the Toronto region, workers had an aberrant heterozygous banding pattern. PGI is variable in most species, and here again most electromorphs have been observed in some species but not in others.

Harpagoxenus Harpagoxenus sublaevis and H. canadensis are indistinguishable from each other in all studied enzymes, and all their electromorphs are also present either in one of their hosts (L. acervorum, L. muscorum and L. gredleriin H. sublaevis and L. spp. A and B in BIOCHEMICAL STUDIES ON PARASITIC ANTS AND HOSTS 199

TABLE 2. FREQUENCY OF DIFFERENT ELECTROMORPHS IN POPULATIONS OF LEPTOTHORAX (S.STR.), DORONOMYRMEX, AND HARPAGOXENUS

MDH-1 IDH PGD ~I v s f x s f v s f x u v s m f x

L. acervorum Reichswald, D X -- -- X -- X X -- × X X Oberleinach, D X -- -- X -- X X Babenhausen, D X -- -- X -- X X -- X X X X Nyehusen, S X -- -- X -- X X X X X La Villette, F X -- -- X -- X X Gran Sasso, I x -- -- x x -- -- x -- X Hokkaido, J x -- -- x -- X X Denali, N. P., Alas. x -- -- x -- X X L. muscorum Reichswald, D X * -- X -- X * -- X X Nyehusen, S X -- -- X -- X X La Villette, F X -- -- X -- X X X Akkus, TR X -- -- X -- X X L. gred/eri Sommerhausen, D X -- -- X -- X X L. retractus Rouyn-Norenda, Que. X -- X -- X -- X Maligne C.; Alta. X -- X -- X -- X Kananaskis, Alta. X -- X -- X -- -- X X L. crassipilis Curecanti, Colo. X -- -- X -- X X L. sp.A Tadoussac, Qu@. X -- -- X -- )4 -- X X Rouyn-Noranda, Qu@. X -- -- X -- X -- X X X Mt. Monadnock, N. H. X -- -- X -- X X L. sp.B Tadoussac, Qu@ -- X X -- X -- X X -- York, Ont. -- X X -- X -- X X -- Milton, Ont. -- X X -- X -- X X -- Moncton, N. B. -- X X -- X -- X -- Mt, Monadnock, N. H. -- X X -- X -- X X -- L. sp.C Maligne C., Alta. X -- * X -- X X Banff, Alta. X -- -- X -- X X L. sp.D Maligne Lake, Alta. -- X X -- X -- X L. paraxenus Milton, Ont. X -- -- X -- X X L. wilsoni Moncton, N. B. X -- D. pocahontas Maligne, C., Alta. X -- -- X X X D. pacis La Villette, F X -- -- X D. goessvvaldi La Villette, F X -- -- X -- X X X -- D. kutteri Nyehusen, S X -- -- X -- X X H. sublaevis Reichswald, D X -- )'< -- X -- X FL canadensis St. Sim@on, Qu@. X -- X -- X -- X Tadoussac, Que. X -- X -- X -- X F. quebecensis Waswanipi, Qua. X X -- X X

Electromorphs are characterized by their different electrophoretic velocities (u = extremely slow, v : very slow, s = slow, m = medium, f = fast, x = very fast). More detailed data on enzyme patterns and collecting sites of non-parasitic Leptothorex (s.str.) have been published elsewhere [17]. (* Elect~morph is very rare.) 200 J. HEINZE

H. canadensis) or other species belonging to the subgenus Leptothorax (s.str.). From the present data it is not possible to confirm a closer relationship between H. sublaevis and H. canadensis to L. acervorum which had been suggested on morphological criteria and esterase banding patterns [19]. In fact, the only Leptothorax (s.str.) species which is similar to Harpagoxenus in all studied enzymes, L. retractus, is the only species which is not parasitized by the slave-makers. It appears likely that Harpa- goxenus diverged at an early stage of speciation from the ancestors of Leptothorax (s.str.) and radiated secondarily over several host species. Harpagoxenus americanus, on the other hand, differs in all enzymes which could be stained in the little material available--MDH, IDH, ME, PGD, TO and esterases--from its presumed congeners and from Leptothorax (s.str.). It shares some electromorphs with its hosts, L. (Myrafant) Iongispinosus, L. (114.) ambiguus, and L. (M.) curvispinosus and other North American Myrafant, e.g. MDH-1 (Fig. 1). Though the data base is small, it supports the idea that H. americanus is not closely related to the other species of Harpagoxenus and to Leptothorax (s.str.) but instead attached to Myrafant. Given the comparatively high genetic variability in the only extensively studied North American Myrafant, L. (M.) Iongispinosus [14], and the little knowledge we have of other nearctic species, it is not yet possible to decide whether H. americanus is closer to one of its hosts than to any other non-parasitic species. Nevertheless, the data corroborate recent proposals to split the genus Harpagoxenus. It had been concluded from charac- teristics such as sexual behaviour, host specificity, and ecological needs that H. americanus does not belong to H. sublaevis and H. canadensis [20, 23]. Though a formal revision has not yet been published, several authors have de facto transferred H. americanus to its own taxon, Protomognathus [1, 4].

Doronomyrmex The European workerless parasites Doronomyrmex goesswald/; D. kutter~ and D. pacis are morphologically similar to their common host, L. acervorum. In all four species, antennae and legs are covered with abundant suberect hairs, which are lack- ing in the other non-parasitic palaearctic Leptothorax (s.str.). D. kutteri and D. goesswa/di had originally been described as species of Leptothorax. In the laboratory, it was possible to cross D. pacis or D. goesswaldi males with D. kutteri females and viable hybrid offspring were produced [20, 24]. Therefore, and because of common karyological and morphological characters, it was concluded that the three species belong to one single genus. All enzyme electromorphs seen in the palaearctic Doronomyrmexare also common in L. acervorum, with the exception of the very slowly migrating PGI-electromorph of D. pacis, which has as yet been found only in a single I. acervorum nest from central Italy (Table 2). D. goesswaldiand D. kutteriare also indistinguishable from their host in esterase patterns (no data available for D. pacis) [19]. In one of three D. goesswa/di colonies, a female had a heterozygous PGI banding pattern. This was the 9nly heterozygous banding pattern observed in a workerless parasite. Though only few nests from single populations could be studied, it is likely that a more extensive search will confirm low heterozygosity levels in these species. Workerless parasites are rare and females mate near or within the nest, thus the inbreeding coefficient should be high. Though at least D. goesswaldiand D. kutteri have electromorphs identical to their host, L. acervorum, the data do not suffice to decide whether Doronomyrmex evolved directly from the host species, as the strong version of Emery's rule and the models of some authors predict [5, 25, 26]. The two species are also biochemically very similar to a second, not parasitized species, L. muscorum, from which they differ in their esterase banding pattern [19]. The genetics of this complex set of 20 bands are not clearly understood and for an estimation of phylogeny this system so far is not appropriate. Two more L eptothorax (s.str.) species differ little from Doronomyrmex: L. gredler/has Q O

-I-

FIG. 1. MALATE DEHYDROGENASE(MDH) PATTERNSOF ADULT LEPTOTHORACINEANTS, SEPARATEDBY tEF ON 0.2 MM THIN POLYACRYLAMIDEGELS (pH RANGE 4-9, APPROX. 4500 Vh). a-c: L. (Myrafant)ambiguus (a from Shawinigan, Quebec, b and c from Burlington, Vt.), d: L. (M.) curvispinosus(Cambridge, Mass.), e: Flarpagoxenusamericanus (Renselaerville, N. Y.), f-h: Harpagoxenus canadensis(f and g from Rouyn-Noranda,Que., h from Bale Ste. Catherine, Que.), i: L. (M.) rugatulus(Ogden, Ut.), j-n: L. sp.A (from La Baie, Shawinigan,Ashuapmushuan, Matagami, and Tadoussac, all in Quebec),o: L. sp.B (Trois Pistoles,Que.), p: L. gredleri(Sommerhausen,Germany), q: L. sp.A (Bar Harbor, Me), r: Formicoxenusquebecensis (Waswanipi, Que.), s: hybrid of L. spp.A and B (Chapais,Que.), t and u: L. muscorum(Nyehusen, Sweden, and La Villette, France) v: L. retractus (Aiguebelle, Que.), w: L. acervorum(Pipplinger Au, Germany). The allozymes of the more cathodal MDH-2 could not be separatedin this pH gradient. Arrows indicate the position of tetrazoliumoxidase (TO)electromorphs, visible only as lighter bands against the backgroundof the gel, which is slightly violet due to reduced tetrazolium salts The lines between lanes i and j, and between r and s indicate the position of water-soluble,coloured pl marker proteins, which were lost during the histochemical staining.