Jürgen Heinze.
The origin of workerless parasites in Leptothorax (s. str.) (Hymenoptera: Formicidae).
Psyche 102:195-214, 1995.

Full text (searchable PDF, 1768K)
Durable link: http://psyche.entclub.org/102/102-195.html

The following unprocessed text is extracted from the PDF file, and is likely to be both incomplete and full of errors. Please consult the PDF file for the complete article.

THE ORIGIN OF WORKERLESS PARASITES IN
LEPTOTHORAX (SSTR.) (HYMENOPTERA: FORMICIDAE) Theodor-Boveri-Institut (Biozentrum der Universitat), LS Verhaltensphysiologie und Soziobiologie, Am Hubland, D-97074 Wurzburg, F.R.G.
The evolutionary origin of workerless parasitic ants parasitizing colonies of Leptothorax (s.str.1 is investigated using data on mor- phology, chromosome number, and allozyme phenotype of both social parasites and their hosts. Of the three previously proposed pathways, the evolution of workerless parasites from guest ants or slave-makers is unlikely, at least according to a phenogram obtained by UPGMA clustering of Nei's similarities based on seven enzymes. Intraspecific evolution of the workerless parasites Doronomyrmex goesswaldi, D. kutteri, and D. pacis from their common host, Leptothorax acervorum cannot be excluded with the present data. The workerless parasite L. paraxenus, however, clearly differs from its host, L. cf. canadensis, in morphology and biochemistry, and most probably did not evolve from the latter species. It is proposed to synonymize Doronomyrmex under Lep- tothorax (s. str.).
Eusocial insects by definition are characterized by a division of labor between non-reproductive workers and reproductive queens. Nevertheless, in a small minority of ant, bee, and wasp species, the worker caste has been secondarily lost. Instead of founding their own colonies solitarily, the queens of these workerless social para- sites invade the nests of other, often closely related host species and exploit the present worker force to rear their own young. In 'present address: Zool. Inst. I, Univ. Erlangen-Numberg, Staudtstrasse 5, D-91058 Erlangen, Gemany
Manuscript received 19 Junuuiy 1996.
195




================================================================================

196 Psyche [VOI. 102
ants, the queens of some parasite species kill all host queens ("murder parasites," Faber, 19691, but in other species parasite and host queen live and reproduce together (inquilines in the strict sense, e.g., Bourke and Franks, 1991). The evolution of social par- asites and in particular of socially parasitic ants has been exten- sively discussed starting with Darwin (1859). Three main routes leading to workerless parasitism have been proposed: workerless parasites might evolve a) directly from the species or species group serving them as host (Emery, 1909; Wasmann, 1909; Kutter, 1969; Buschinger, 1990; Bourke and Franks, 1991); b) from other para- sites, such as temporary parasites, slave-makers, or guest-ants (Wasmann, 1908, 1909; Emery, 1909; Wilson, 1971); or c) from non-parasitic ancestors other than the host species (West-Eberhard, 1990; Bourke and Franks, 199 1).
The myrmicine tribe Formicoxenini (formerly Leptothoracini, Bolton, 1994) is extraordinarily rich in social parasites and thus provides an ideal system to investigate the evolutionary pathways to workerless parasitism (Buschinger, 1986, 1989, 1990). Lep- tothorax (s-str.) (i.e, L. acervorum, L. muscorum, L. cf. canadensis and several other non-parasitic taxa), Formicoxenus, Harpagox- enus, and the palaearctic Doronomyrmex appear to be especially closely related (Buschinger, 198 1, 1987) and have been grouped in a distinct subtribe within the Formicoxenini (Loiselle, Francoeur and Buschinger, 1990). They nevertheless exhibit remarkably dif- ferent life histories. Formicoxenus are guest-ants, which live in the nests of Formica, Myrmica, or Manica (Francoeur et al., 1985). Formicoxenus workers beg food from their hosts but rear their own brood in separate chambers close to the host. The host colonies remain intact and continue to produce sexual brood (e.g., Wheeler, 1910). Harpagoxenus are slave-makers, whose queens after invad- ing a Leptothorax (s-str.) host colony kill or expel all adult resi- dents. Leptothorax workers which eclose from the conquered brood serve as "slaves" and take care of the slave-maker queen's larvae. Harpagoxenus workers eventually pillage brood from neighboring Leptothorax nests, which after eclosion serve as addi- tional slaves. Doronomyrmex kutteri and D. pacis are workerless parasites which tolerate the Leptothorax host queens, though they probably decrease host reproductive success by feeding on their eggs (Kutter, 1969; Franks et al., 1990). D. goesswaldi, L. para- xenus, and L. wilsoni are workerless parasites which kill the host



================================================================================

19951 Heinze 197
queen, but not the adult host workers (Buschinger and Klump, 1988; Heinze, 1989; Heinze and Alloway7 1991). Several attempts have been made to deduce the evolutionary ori- gin of formicoxenine murder parasites and inquilines from data on morphology, karyotype, and enzyme phenotype (Buschinger, 198 1, 1990; Heinze? 199 1). Reviewing these previously published results and providing additional unpublished data, here I critically exam- ine the hypotheses on the evolution of workerless parasites in this group and provide evidence that routes a and c are the most likely pathways leading to workerlessness.
Colonies of parasitic and non-parasitic Formicoxenini (Table 1) were collected during the last 10 years in various parts of North America, Europe, and Turkey. Ants used in this study are from the following sites. Parasites: L. paraxenus: Bic (Co. de Rimouski, Quebec), Milton (Halton Co., Ontario); L. wilsoni: Mt. Monadnock (Cheshire Co., New Hampshire), Escoumins (Co. de Saguenay, Quebec)? Jasper Nat. Park (Alberta; Buschinger and Schumann, 1994); Doronomyrmex goesswaldi: La Villette (Dept. Hautes- Alpes, France); D. kutteri: Leinburg (Bavaria? Germany); D. pacis: La Villette (Dept. Hautes-Alpes, France); Harpagoxenus sublaevis: Rudolstadt (Thuringia, Germany); Formicoxenus quebecensis: Waswanipi (Co. de Abitibi,
Mt. du Lac des Cygnes (Co.
de Charlevoix-Est, Quebec)? Jasper Nat. Park (Alberta; Buschinger, Schumann and Heinze, 1994). Non-parasitic species: L. sp. A: Tadoussac (Co. de Saguenay? Quebec); L. acervorum: Grossos- theim (Bavaria, Germany), Leinburg (Bavaria, Germany), Ilgaz Dagi Ge~idi (Cankiri? Turkey); L. cf. canadensis: Bic (Co. de Rimouski? Quebec)? Tadoussac (Co. de Saguenay, Quebec), Mount Monadnock (Cheshire Co., New Hampshire); L. gredleri: Sommer- hausen (Bavaria? Germany); L. muscorum: Leinburg (Bavaria, Ger- many)? Ilgaz Dagi Ge~idi (Cankiri? Turkey); L. ''muscorum~' C: Maligne Canyon (Alberta); L. retractus: St. Simeon (Co. de Charlevoix-Est); L. sphagnicolus: L'Ascension (Co. de Chicoutimi, Quebec); D. pocahontas: Maligne Canyon (Alberta). Details on the collecting procedure, laboratory rearing, and the life histories and collecting sites are published elsewhere (Heinze, 1989, 1993; Heinze and Ortius, 1991; Heinze, Trunzer, Lechner and Ortius, 1995).




================================================================================

198 Psyche [vo~. 102
Table 1. Synopsis of guest-ants, slave-makers, murder parasites, and queen-toler- ant inquilines thought to be related to the ant subgenus Leptothorax (sstr,). Geographical
Species TY ~e Host species range References Doronomyrmex murder
goesswaldi parasite
D. kutteri inquiline
D. pacis inquiline
L. faberi murder
parasite?
L. paraxenus murder
parasite
L. wilsoni murder
L. acervorum Alps
L. acervorum Alps, S. Sweden,
Estonia
L. acervorum Alps
L. cf. canadensis Maligne Lake,
Aha.
L. cf. canadensis Ontario, Qukbec
L. cf. canadensis, Qukbec, New
Buschinger and
Klump, 1988
Buschinger,
1971
Buschinger,
1971
Buschinger,
1982
Heinze and
Alloway, 199 1
Heinze, 1989,
parasite L. sp. A
Harpagoxenus slave-maker L. acervorum,
sublaevis and L. canadensis,
H. canadensis L. gredleri,
L. muscorum,
L. sp. A
Formicoxenus guest-ant Myrmica spp.,
SPP. Formica spp.,
Manica mutica
Brunswick, New Heinze et al.
Hampshire, Rocky 1995,
Mountains Buschinger and
Shumann, 1994
coniferous forests Buschinger,
in Northeastern
197 1 ; Heinze,
North America Stuart,
and Eurasia Alloway, and
Buschinger,
1992, Heinze
and
Kauffmann,
1993
holarctic Francoeur et
al., 1985
L. cf. canadensis, the most common Leptothorax (s-str.) in New England, Qukbec and the Canadian Maritime Provinces is also referred to as ''large black L. 'muscorum~~' (e.g., Francoeur, 1986; Loiselle et al., 1990) or Leptothorax sp. B (e.g., Heinze and Buschinger, 1987, 1988; Heinze, 1989). However, the original description of L. canadensis (Provancher, 1887) fits quite nicely to this taxon. At present it is not known how far west L. cf. canaden- sis ranges, but morphologically, karyologically, and biochemically more or less similar ants (referred to as Leptothorax ''muscorum~~ D and E, Heinze, 1989) occur throughout the Rocky Mountains and the Coast Mountains in western Canada and the western USA. Leptothorax sp. A is a widespread species in open coniferous forests and on partly shaded rocky patches in New England,



================================================================================

19951 Heinze 199
Qukbec, Ontario, and New Brunswick; a morphologically similar species, though with a different chromosome number, Leptothorax "muscorum~~ C, perhaps identical to the variety L. muscorum septentt-ionah (Wheeler, 19 171, is found in the Canadian Rocky Mountains.
Chromosomes were prepared from unpigmented male Leptotho- rax pupae following a procedure by Imai, Crozier, and Taylor 1977; (see also Loiselle, et al, 1990). For electrophoresis in 12.5 cm long 7.5% polyacrylamide gels, adults or pupae were crushed individually in 40pl of PAGE-homogenization buffer (0.lM TrislHC1 pH 8.0, 1mM EDTA, 0.05mM NADP, 2mM p-Mercap- toethanole, 10% glycerine, 0.01 % bromothymol blue), of which 5 to lop1 were applied to the gel (gel buffer: 0-125M TrisfHCl pH 8.0; tray buffer 0.16M glycine, 0.025 M Tris, pH 8.3). Proteins were separated at 10å¡ with 10mA per gel for 1.5 hours. For detection of IDH, 0.25M TrislHCl pH 9.6 was used both in the gel and as tray buffer. For electrophoresis on cellulose acetate plates (Titan 111, Helena Laboratories, Beaumont, Texas), whole ants were crushed in 5-8pl tray buffer (with 0.01% bromothymol blue and amaranth as tracking dyes) and applied to the surface of the pre-soaked gel (tray buffer: 0.1M TrisfO. 1M MaleatlO.0lM EDTA, pH 7.4, 1 : 10) using the Helena "Super Z" applicator. Gels were run for 30 min. at 200 V at 5OC (see also Heinze, 1991). Of 14 to 20 enzymes screened in Leptothorax acervorum, L. cf. canadensis, and L. sp. A, 7 which could reliably be stained and showed a rea- sonable amount of inter- or intraspecific variation, were chosen for a more detailed analysis of both host species and parasites. In most non-parasitic species, at least 20 workers from 10 different colonies were stained for each enzyme; sample sizes are typically much larger in GPI and PGM. Fewer workers were available from L. sphagnicolus, L. retractus, and the parasitic taxa. In species with limited material, where no allozyme differences were found between populations (L. paraxenus, L. wilsoni, I? quebecensis), data from different populations were pooled for the analysis. Not all of those enzymes found to be invariable among the non-para- sitic species were stained in the workerless parasites. Nei's indices were calculated from allele frequencies and clustered (UPGMA) using the computer program NTSYS (Rohlf, 1990); in addition, a neighbor-joining tree (Saitou and Nei, 1987) was calculated. Stability of clusters in the UPGMA phenogram was tested by



================================================================================

200 Psyche [vo~. 102
jackknifing over taxa (Lanyon, 1985) and by calculating a cophe- netic regression coefficient.
Voucher specimens, wherever available, of the studied species are deposited in the MCZ, Cambridge, Mass., (USA). Morphology
The current taxonomic confusion concerning the nearctic repre- sentatives of Leptothorax (s.str.) (e.g., Creighton, 1950; Brown, 1955; Francoeur et al., 1985; Heinze, 1989) makes a thorough morphological comparison between non-parasitic Leptothorax (s.str.) and the associated parasitic taxa difficult. Only few charac- ters appear to be stable enough to serve for species distinction (Table 2). The most reliable and most frequently cited characters are the suberect hairs on tibia and scapes in L. acervorum, L. sphagnicolus, Doronomyrmex, and Harpagoxenus, and the indented clypeus in L. gredleri, L. retractus, L. paraxenus, and to a lesser extent also Leptothorax sp. A (Table 2). Both characters, however, may vary between populations: L. acervorum from Alaska, for example, are considerably less hairy than central Euro- pean specimens (Heinze and Ortius, 1991). Other characters which serve to distinguish parasitic genera from Leptothorax (s. str.) are probably mostly adaptations to the parasites' specialized way of life and cannot be used to investigate phylogenetic relationships. Harpagoxenus, e.g., is easily recognized by strong, toothless mandibles and an extraordinarily large head with antenna1 scrobes, all obvious adaptations to slave-making. Similarly, the morpholog- ical features separating Doronomyrmex from Leptothorax are thought to be adaptations to the parasitic life, and some authors question whether Doronomyrmex should be kept as a distinct genus (Brown, 1973; Bolton, 1982).
In a detailed morphological revision, Francoeur et al. (1985) provided clear evidence for the separation of the guest-ant genus Formicoxenus from non-parasitic Leptothorax (s.str.). Female Formicoxenus are more slender than Leptothorax and the scape of Formicoxenus males is longer and more cylindrical than in Lep- tothorax males. Furthermore, whereas the majority of Formicox- enini, including non-parasitic Leptothorax and Harpagoxenus,



================================================================================

19951 Heinze 201
Table 2. Morphological characteristics and chromosome numbers of Leptothorax (s.str.) and their parasites, based on Bolton (1982), Francoeur et al. (1985), Loiselle et al. (1990), and own observations.
Eyes Erect Enlarged
with hairs on Dufour's Clypeus Palp
hairs legs gland indented formula Chromosomes L. acervorum
L. cf. canadensis
L. gredleri
L. muscorum
L. retractus
L. sphagnicolus
L. sp. A
L. c
D. pocahontas
L. faberi
L. paraxenus
L. wilsoni
D. goesswaldi
D. kutteri
D. pacis
H. canadensis
H. sublaevis
Formicoxenus spp.
5,3
593
5,3
5,3
5,3
5,3
5,3
5,3
593
5,3
5,3
4,3
5,3
4,3 and 5,3
5,3
5 93
4,3 or 5,3
have a standard palp formula of 5 maxillary palps and 3 labial palps, the number may be reduced to 4, 3 in Formicoxenus (Bolton, 1982). However, as shown by Francoeur et al. (1985), there is intergeneric and even intraspecific variation in palp formula in Formicoxenus. The palp formula is similarly reduced to 4, 3 in Leptothorax wilsoni (two queens from Jasper N.P., Alberta, and one individual from Escoumins, Quebec-the palp formula given in Heinze, 1989 for the type material needs to be confirmed by re- examination of material from the type locality) and Doronomyrmex pacis (two individuals from Jenner, Germany; according to Kutter, 1950 males of D. pads from Switzerland have a palp formula of 5, 4). Palp formula is therefore probably not very informative on genus level (Table 2).




================================================================================

202 Psyche [vo~. 102
Formicoxenus have scattered hairs on their compound eyes, whereas the eyes are thought to be hairless in Leptothorax (Fran- coeur et al., 1985). Hairy eyes, however, are found in L. wilsoni (Heinze, 1989; Table 2). Despite the superficial similarity between L. wilsoni and Formicoxenus in these two characters, L. wilsoni is nevertheless morphologically closer to Leptothorax in others, such as the scape length in males. L. wilsoni shares reduced mandibular dentition with Epimyrma, a slave-making satellite genus of Lep- tothorax (Myrafant), but clearly differs in palp formula and the shapes of petiole and postpetiole.
Most queens of workerless parasites in Formicoxenini, and per- haps of workerless parasitic ants in general, are extraordinarily small compared to queens of the host species (Douwes, 1990; Nonacs and Tobin, 1992). Queens of L. paraxenus are a notable exception, in that they are of similar size as the host queens. As small size is thought to be adaptive in parasites-parasite queens do not need much resources for colony founding and thus, a larger number of less well equipped, small queens can be produced from the same amount of energy available to the colony (Douwes, 1990)-the condition in L. paraxenus might reflect a recent origin from a non-parasitic ancestor.
Parasite queens are typically characterized by a broadened post- petiole, a strong ventral petiolar spine, and a larger Dufour's gland compared to their hosts. These features, however, are not restricted to social parasites but may be found to a varying degree also in the queens of free-living species such as Leptothorax sp. A and L. gredleri.
Chromosome Number and Allozyme Phenotype The significance of chromosome number as a taxonomic charac- ter is very poorly understood. Nevertheless, various studies have used chromosome analyses to clarify the taxonomy of Formicox- enini (Fischer, 1987; Heinze and Buschinger, 1989; Loiselle et al., 1990; Buschinger and Fischer, 199 1). Chromosome numbers in Leptothorax (s.str.) and associated genera range between 11 and appr. 28. Palaearctic Doronomyrmex have much higher chromo- some numbers than their common host, L. acervorum. In contrast, the nearctic workerless parasites L. faberi (Buschinger, 1982) and L. paraxenus both have 15 chromosomes (L. paraxenus: 30 metaphase plates from 4 male pupae from Milton, Ontario),



================================================================================

19951 Heinze 203
whereas their host L. cf. canadensis has 17 or 18 (Heinze and Buschinger, 1989; Loiselle et al., 1990). UPGMA clustering of Nei's indices calculated from seven enzyme systems which are of diagnostic value in the studied species results in the phenogram shown in Fig. 1. Original data and Nei's indices are given in Tables 3 and 4. Goodness of fit of the cluster analysis to the data was tested by comparing a matrix of cophenetic values, calculated from the tree matrix, with the origi- nal similarity matrix. The resulting cophenetic correlation of r = 0.809 suggests a rather mediocre fit (Mantel t-test, t = 6.53, p = 1.000). Nevertheless, the overall branching pattern is remarkably stable. Two fundamental dichotomies are found in all 21 Fig. 1. Phenogram obtained by UPGMA clustering of Nei's indices from electro- morph frequencies of several species of Leptothorax (s.str.), Doronomyrmex, Formi- coxenus quebecensis, and Harpagoxenus sublaevis. For abbreviations see Table 3.



================================================================================

204 Psyche [vol. 102
pseudoreplicates generated from the original data set by jackknif- ing over taxa. Firstly, Formicoxenus, represented in the elec- trophoretical study by F, quebecensis, is least similar to the other taxa, and secondly, a cluster consisting of Harpugoxenus sublaeuis, L. retractus, L. cf. canadensis, and L. wilsoni (cluster A), stands in sharp constrast to the remaining taxa (L. acervorum, L. sphagnico- lus, L. muscorum, L. gredleri, L. paraxenus and Doronomyrmex, cluster B). L, acervorum, L. sphugnicolus and the two workerless parasites, D. kutteri and D. goesswaldi, also form a stable group (supported in all pseudorepHcates), which is probably reflected in morphological similarities among these species, such as the pres- ence of erect hairs on the scapes and legs. Though Harpagoxenus and D. pads likewise share this character, at least Harpagoxenus appears biochemically quite dissimilar from the L. acervorum group. The branching pattern in the lower half of cluster B (L. muscorum, L. gredleri, L. sp. A, L. paraxenus, D. pacts, and D. pocahontas} was present in only 19 of 21 replicates and thus is less well supported by the data. It therefore would be premature to con- clude that D. pads is not very close to its workeriess congeners Table 3. Frequency of electromorphs of 7 enzymes in rite ant genus LeptMhorax kstr.) and associated parasitic taxa. The abbreviations stand for the followinc .,
species: Dgoe: 0. goesswaldi (9 virgin queens from 3 colonies); Dku: D. kutieri (8 virgin queens from 2 colonies); Dpac: D. pacis (5 virgin queens from 3 colonies); Dpoc: D. pocahontax (10 workers from 3 colonies); Fqu: Formicoxenus quebecewis (12 workers from 4 colonies from different populations); Hx: Harpapxenus sublae- vis (26 workers from 8 colonies); LA; L. sp. A (209 workers from 17 colonies; LaG, LaR, LaT: L. acervorum from Gropostheirn (Germany, 196 workers from 89 colonies), Reichswald (Germany, between 64 and 1420 workers from 10 to 140 colonies), and Ilgaz Dagi Gyidi (Turkey, 49 workers from 7 colonies); LC: Lep- tothorax C (10 workers from 4 colonies); LcB, kM, LcT: L. cf. canadensis from Bic (Quebec, ! 13 workers from 13 colonies), Mt. Monadnock (New Hampshire, 131 workers from 17 colonies) and Tadoussac (Qud~w. 66 workers from 10 colonies); Lgr: L. gredleri (between 20 and 299 workers from 29 colonies); hu, LmuT, L. muscorum from Reichswald (between 20 and 569 workers front 76 colonies) and Ilgaz Dagi Ge~idi (20 workers from 5 colonies); Lpm L. paraxenus (10 virgin queens from 4 colonies}; Let: L retractus (10 workers from 2 colonies); Lsp: L. sphagnicolus (10 workers from 2 colonies); h i : L witsod (8 virgin queens from 4 colonies).
The enzymes are Glucose-6-phosphate isomerase (GPI, referred to as PGI in Hei me, 1989, 199 1). Phosphoglucomutase (EM), 6-Phosphogluconate dehydmge- nase (PGD), Malate dehydrogenase (MDH), Ismitrate dehydrogenase (IDH), Matic enzyme (ME), and Lactate dehydrogenase (LDH). All species share the same elm- tromorph of tetrazolium oxidase. v, s, rn, n, f, and x denote different migration velocity during electrophoresis; blanks in the table indicate missing data.



================================================================================




================================================================================

Table 4. Nei's indices calculated from electromorph frequencies in Table 3. - - - - - - pp
Dgoe Dku Dpac Dpoc Fqu HxLA LaG LaR L a T L C LcB LcM LcT Lgr Lmu LmuT Lpar Lret Lsp Lwi Dgoe 0.000
Dku 0.044 0.000
Dpac 0.221 0.288 0.000
Dpoc 0.381 0.582 0.288 0.000
Fqu 1.055 0.847 0.693 1.925 0.000