Morphometrics , life history and population biology of the Ponto-Caucasian slave-making ant Myrmoxenus tamarae ( Hymenoptera : Formicidae )

The ant genus Myrmoxenus consists of about ten socially parasitic species including active slave-makers and workerless “degenerate slave-makers”. Myrmoxenus tamarae was previously known only from type material, two workers collected at Daba, Georgia and nothing was known about its life history, colony structure or the morphology of its sexuals. An inspection of colonies of M. tamarae near the type locality in 2010 indicates that young queens of M. tamarae invade Temnothorax nests and kill the host queen by throttling. The simultaneous presence of two slave species in a single colony (an undescribed species related to T. nylanderi and a species morphologically resembling T. unifasciatus) indicates that M. tamarae is an active slave-maker. The genetic structure of the colonies matches that expected for a monogynous and monandrous ant, but three of eight colonies inspected appeared to contain workers belonging to an additional genetic lineage.


INTRODUCTION
Ants are indisputedly one of the big successes in evolution.Their complex cooperative life enables them to thrive in numerous ecological niches (Wilson, 1971;Hölldobler & Wilson, 1990, 2008), but at the same time makes them particularly vulnerable to exploitation by pathogens or parasites (Schmid-Hempel, 1998).Among the latter are slave-making ants ("dulotic ants";Wilson, 1975;Buschinger, 1986Buschinger, , 2009;;Hölldobler & Wilson 1990;d'Ettorre & Heinze, 2001).Instead of founding new societies independently, young slave-maker queens usurp the nests of other ant species, in which they kill or expel the resident queen(s) and often also adult workers.Host workers that later emerge in the conquered nest misidentify the slavemaker queen as their own (e.g., Lenoir et al., 2001) and, as in unparasitized nests, take over all the daily duties in the colony.Workers produced by the slave-maker queen herself do not engage in brood care or foraging but instead pillage pupae from host nests in the neighborhood to replenish the stock of slaves "at home" (Wilson, 1971;Buschinger, 1986Buschinger, , 2009;;Hölldobler & Wilson, 1990;d'Ettorre & Heinze, 2001).
Myrmoxenus Ruzsky, 1902 is the most species-rich of six or more clades of slave-making ants that have evolved in the myrmicine tribe Formicoxenini (Beibl et al., 2005).It comprises 10 to 12 species in Southern Europe and Northern Africa (Buschinger, 1989;Schulz & Sanetra, 2002).Two workers collected in 1963in Georgia (Jijilashvili, 1967) were described as M. tamarae (Arnol'di, 1968), but since then no additional specimens of this taxon have been collected.Here we report on the results of two field trips to the type locality of M. tamarae near Daba, Borjomi district, Georgia, during which we collected a total of nine colonies of this rare and probably endangered species of ant.We describe in detail the habitat of this ant, morphology of its sexuals and social and genetic composition of its colonies.A multidisciplinary comparison of M. tama rae and other species of Myrmoxenus does not support the view that M. tamarae is a junior synonym of M. ravouxi (André, 1896) and indicate that it can be considered as a valid species (Gratiashvili et al., 2014).

Ant collecting and culture
Colonies of M. tamarae and its two host species were collected in August 2010 (n = 5) and June 2011 (n = 4) from their nests under pine bark and in rock crevices at the type locality near Daba, Borjomi Gorge (Borjomi district, South Georgia) (Fig. 1a).The structure of six colonies was investigated immediately after sampling.Then the colonies were transferred to laboratories in Tbilisi and Regensburg and kept in standard three-chambered plastic boxes with a plaster floor under semi-natural conditions as described previously for other formicoxenine ants (Buschinger, 1974;Heinze & Ortius, 1991).

Morphometry
We measured 21 morphological characters of each of 20 mounted and dried workers (from six nests), five queens (incl.female sexuals, from two nests), and five males (from one nest) using a pin-holding stage, permitting unlimited rotation around the X, Y and Z axes, and a Wild M 10 stereomicroscope equipped with a 1.6 × plan-apochromatic objective at magnifications between 200 × and 320 ×.Measurements are defined by Seifert (2006) and Gratiashvili et al. (2014).Automontage pictures were produced using a Keyence VHX 500 FD digital microscope.tested against the value expected for workers in monogynous and monandrous ant colonies (0.75) using t-test.

RESULTS
The type locality of M. tamarae is on the south-eastern slope of the Gvirgvina mountains above the right bank of Gujaretistskali river (Fig. 1a) near the cemetery at Daba (Borjomi district, Southern Georgia; 41°48.5´N,43°27´E; 1000-1016 m a.s.l.).The habitat is mixed forest with Scots pine (Pinus sylvestris var.hamata) and, less abundant, Caucasian spruce (Picea orientalis) and a xerophyte herbaceous plant cover ("Pineta xeroherbosa": Akhalkatshi & Tarkhnishvili, 2012; complete list of vegetation in Table 1).Numerous nests of two species of Temnothorax Mayr, 1861 were found under the uppermost bark layers at the base of living pines and in rock crevices.Approximately 8% of the Temnothorax colonies inspected were parasitized, i.e., contained workers and / or queens of M. tamarae.
We determined the colony structure of M. tamarae by genotyping workers at loci 2MS60, 2MS65 (Suefuji et al., 2011) and 2MS86II (forward primer: AGA TTC ACT TGA AGC AGG AGC; reverse primer: ACG CTT GGG CAT AAT AAA AAA TC).Standard population genetic parameters (number of alleles, expected and observed heterozygosities, frequency of null alleles) were determined using the software Cervus 3.03 (Kali nowski et al., 2007).Worker relatedness was estimated following Queller & Goodnight (1989) using GenAlEx (Genetic Analysis in Excel, an Excel add-in: Peakall & Smouse, 2012).Standard errors were obtained by jackknifing over loci.Mean relatedness estimates were Table 2. Means of morphometric data of samples of workers, queens and males of Myrmoxenus tamarae.Data are arithmetic means ± standard deviations.Upper and lower extremes are given in square brackets.n is the number of colonies sampled, i the total number of individuals studied in these samples.Characters were measured as defined by Seifert (2006) and Gratiashvili et al. (2014).

Abbreviation
Definition of measurements and ratios Workers (n = 6; i = 20 ) Queens (n = 2; i = 5) and M. tamarae can be distinguished only by complex statistical methods (Gratiashvili et al., 2014).Field data show that at the type locality M. tamarae utilizes as hosts two species of Temnothorax.Six nests contained host workers of a species morphologically identified as T. unifasciatus (Latreille, 1798) by B. Seifert and A. Schulz, but genetically forming the out group to Central European T. unifasciatus and T. nigriceps (Mayr, 1855) (based on a sequence of the mitochondrial gene COI / COII, M. Suefuji, unpubl.).Two nests contained workers of "T.crasecundus" (provisional name of a recognized morphospecies currently being described by B. Seifert and S. Csősz), which is a species closely related to T. nylanderi (Foerster, 1850) and T. crassispinus (Karavaiev, 1926).One colony contained both slave species.In one laboratory colony we observed a young M. tamarae queen killing a "T.crasecundus" host queen by throttling.
The three microsatellite loci studied were sufficiently variable to give a crude estimate of genetic colony structure.Locus 2MS60 had six alleles, 2MS65 seven alleles and 2MS86II four alleles (mean number of alleles: 5.67; Table 3).The frequency of potential null alleles was estimated to be lower than 0.2 at each locus.Mean relatedness, estimated following Queller & Goodnight (1989), was 0.627 ± SE 0.086, which is not significantly different from the value of 0.75 expected for monogyny and monandry (a single, singly mated queen per colony; t = 1.489, p = 0.180).Nevertheless, in three of eight colonies we found two or three individuals with a genotype at one locus that was not compatible with monogyny and monandry.

DISCUSSION
M. tamarae closely resembles other slave-making species of the genus Myrmoxenus, in particular M. ravouxi, in morphology and life history.Workers, and presumably also queens and males, of M. tamarae can be separated from those of M. ravouxi only by an exploratory statistical analysis of multiple morphological characters, as in other pairs of closely related or cryptic species (e.g., Seifert, 2013;Seifert et al., 2013;Csősz et al., 2013).M. tama rae has therefore been considered to be synonymous with M. ravouxi (Buschinger, 1989;Schulz & Sanetra, 2002), but our recent integrative study revealed subtle differences between the two taxa, which do not support the proposed synonymy (Gratiashvili et al., 2014).
The few colonies of M. tamarae collected provide useful information about the life history of this species, which support our view that M. tamarae is a close relative of M. ravouxi (Gratiashvili et al., 2014).One colony contained workers from two host species, which demonstrates that M. tamarae is an active slave-maker like M. ravouxi.At Daba, M. tamarae utilizes a species of Temnothorax, which morphologically resembles T. unifasciatus.The other host is T. "crasecundus", a species closely related to T. nylanderi and T. crassispinus.This is particularly surprising, as species of this taxonomic group, though abundant throughout much of the range of Myrmoxenus, are rarely if ever parasitized (Buschinger, 1989;Delattre et al., 2012Delattre et al., , 2013; but see Mei, 1992).Active slave-raiding may also be the explanation for genotypes of M. tamarae workers that did not match the typical sociogenetic structure of Myrmoxenus, monogyny and monandry (a single, singly mated queen per colony).Similarly aberrant genotypes are recorded in a large fraction of M. ravouxi colonies in Southern Germany.Genotyping the offspring produced by queens under controlled laboratory conditions strongly supported monogyny and monandry and suggested that genetic heterogeneity in the field might result from colony fusion or raiding of neighboring colonies or adoption of stray Myrmoxenus workers (Suefuji & Heinze, 2014).
As is the case for all species of Myrmoxenus, queens of M. tamarae appear to eliminate the host queen by throttling.We observed such an interaction in a colony that had been queenless when collected one year before.Presumably sexuals of both the host and parasite eclosed from brood in the laboratory.Unfortunately we could not determine whether the queens were inseminated.While sexuals of M. ravouxi leave the nest to mate during nuptial flights, those of M. kraussei (Emery, 1915) and other species regularly mate in the nest, which results in considerable heterozygote deficiency (Suefuji & Heinze, 2014).The genetic data, albeit based on a small sample and only three loci, do not support regular inbreeding in M. tamarae.It is therefore unlikely that the M. tamarae queen had been inseminated.
Genetic diversity in the population studied was relatively high and matched that found at other loci in a similarly small number of colonies of M. ravouxi (Suefuji & Heinze, 2014, population Schönhofen; average number of alleles 6.2).This indicates that the population at the type locality does not suffer from inbreeding and might probably be larger than concluded from the difficulties of finding slavemaker colonies.
The type locality at Daba is the only known population of M. tamarae in Georgia, but the finding of morphologically and genetically similar ants in Northern Turkey close to the Georgian border suggests that the species might be more common throughout the Lesser Caucasus and other

Table 3 .
Primer sequences, repeat motives and annealing temperature (T a ) of the three microsatellite loci studied in the slave-making ant Myrmoxenus tamarae.Allele size range, number of alleles (A), expected and observed heterozygosities (H exp , H obs ) were calculated.N indicates the number of individuals that were genotyped (ten per colony).At least the habitat, dry and open mixed forests with pines, firs and oaks, and the two host species are abundant in many places in this area.