Mixed colonies and hybridisation of Messor harvester ant species (Hymenoptera: Formicidae)

Abstract

The Mediterranean harvester ant species Messor minor, M. cf. wasmanni, and M. capitatus can co-occur in the same habitat. In Italian populations, we encountered colonies that contained workers from more than one species as identified via standard morphology, as well as colonies with workers that appeared to be morphologically intermediate between species. This unusual finding required further analysis. We analysed such colonies using microsatellites, mitochondrial DNA and refined morphometrics, and a simple inference key for the colony-level interpretation of data from the three sources combined. We infer that Messor minor and M. cf. wasmanni engage in bidirectional interspecific gene flow. Hybrids between these two species are inferred to produce fertile offspring, which would indicate that barriers to hybridisation do not exist or can be completely overcome. This is unexpected, given that they are non-sister species and broadly sympatric in nature. Our findings also indicate the possible occurrence of hybrid-hybrid crosses, a phenomenon rarely observed in ants. We cautiously interpret the data at hand as in support of the interspecific gene flow considerably shaping the genetic makeup of populations, raising the question about a potential adaptive value of this hybridisation. Messor capitatus mixes with hybrids of the other two species, but we found no indication of hybridisation involving this species. We discuss various hypotheses on the causations of colony mixing and hybridisation in the three Messor species at the proximate and ultimate level.

Appendices

Appendices

Appendix 1 Allocation of species names

Allocation of species names is complicated for European Messor ants because of the lack of a recent taxonomic revision. We applied the worker-morphology based taxonomic descriptions of Messor minor by André (1883) and of M. capitatus by Latreille (1798), and followed the naming tradition of Italian ant taxonomists, e.g. Fabrizio Rigato (pers. comm. 2008).

Some samples from our study system are morphologically identical with multiple specimens from Italian museum collections identified as “Messor wasmanni Krausse 1910”. The corresponding description by Krausse (1910) is not interpretable, however, as it is merely a mention of stridulation differences from other species. The subsequent description by Krausse (1911) is very brief (“in Größe und Färbung dem M. barbarus sich nähernd, indes hinsichtlich des Pronotums quergestreift wie bei M. barbarus minor André”, i.e. ‘in size and colouration approaching M. barbarus, while the pronotum is transversely striated as in M. barbarus minor André’). No type specimens of M. wasmanni could be found (B. Seifert, unpubl.), which would make it desirable to designate a neotype from the type locality (Asuni, Sardinia). However, in Sardinia, four Messor species have been recorded (Baroni Urbani 1971) – M. capitatus, M. minor, M. wasmanni, and M. structor (Latreille 1798), to all of which the character of a transversely striated pronotum applies – complicating a neotype designation. Messor wasmanni might thus be considered as a nomen dubium at some point in the future.

At the same time, the samples mentioned are morphologically identical with the type specimens of Messor concolor Santschi 1927 (see Santschi 1927) that are preserved at the Museo Civico di Storia Naturale, Genova (S. Csősz pers. comm.). By application of ICZN (1999) Code Article 45.5, M. concolor Santschi is the valid name for Messor barbarus subsp. semirufus var. concolor Emery 1908. The pictorial information on type material in Emery (1908) – which also is the type material of M. concolor Santschi – matches the relevant samples of the present study (medium-length setae on gula from anterior margin backwards to level of posterior margin of eye, and much shorter setae posterior from that level). Conspecificity with M. concolor is further supported by the clustering of mitochondrial DNA sequences from the relevant present samples with four sequences deposited at GenBank (DQ074326–DQ074329) from specimens identified as M. concolor by comparison with the type material (Schlick-Steiner et al. 2006).

In conclusion, we refer to those samples as “Messor cf. wasmanni”, thus acknowledging that at present it is uncertain whether the future will bring about that M. wasmanni is considered a nomen dubium, in which case the valid name may turn out to be M. concolor, or whether the name M. wasmanni will be conserved by a neotype designation.

Appendix 2 Morphometric procedure in detail

Morphometric analysis (244 workers) was performed without knowledge of the molecular genetic results. Specimens were selected from each colony after specimens for genetic analysis had been selected, applying the following rationale: Whenever a screening of all workers based on external morphology suggested a heterogeneity beyond what was considered to be normal intraspecific variation – on the basis of experience with intranidal variation in ants, including in the genus Messor – specimens representing the extreme ends of the variation within the colony were chosen.

All morphometric measurements were made on mounted and dried specimens, using a pin-holding stage that permitted full rotations around the X, Y and Z axes. A Wild M10 high-performance stereomicroscope equipped with a 1.6× planapochromatic front lens was used at magnifications of 160–320×. A Schott KL 1500 cold-light source equipped with two flexible, focally mounted light cables providing 30°-inclined light from variable directions, allowed sufficient illumination over the full magnification range and a clear visualization of silhouette lines. Two cold-light sources were used: the external Schott KL 2500 LCD and the coaxial polarised-light illuminator built into the microscope. Depending on what illumination regime was required, the two light sources were used either simultaneously or alternatively. This provided optimum resolution of any tiny structure and microsculpture at the highest magnifications. A Leica cross-scaled ocular micrometer with 120 gradation marks ranging over 65% of the visual field was used. To avoid parallax error, its measuring line was constantly kept vertical within the visual field. Mean measurement errors calculated were ± 0.6 μm for small and well-defined structures such as scape-base width (ScBW; for details, see list below), and ± 1.5 μm for larger structures that are difficult to position, such as cephalic length.

In the following, the 17 morphometric characters measured are listed, and crucial terms used are defined.

CL:

Maximum cephalic length in median line; the head was carefully tilted to the position with the true maximum; excavations of occiput and/or clypeus reduced CL. Surface irregularities due to sculpture, carinae in particular, were considered by averaging between peaks and valleys of sculpture.

CS:

Cephalic size; arithmetic mean of CL and CW.

CW:

Maximum measurable cephalic width, irrespective of whether it occurred in front of, behind or across the eyes.

EL:

Maximum length of eye. All structurally defined ommatidia, pigmented or not, were included.

Fe2L:

Length of midfemur on extensor side; measured from distalmost point of femur to border between femur and trochantellus (see below). Equivalent measurements were possible with view on plane or edge of femur and intermediate viewing positions.

LNdCV:

Natural logarithm of mean carinae distance on central vertex in μm; measured in dorsal view along a transversal line at level of eye centres and between imagined caudal prolongations of frontal carinae. The number of carinae/carinulae crossing this line was counted; those just touching the measuring line and those exactly at its endpoints were counted as 0.5.

Mesosomal longitudinal axis:

From centre of metapleural gland orifice to lowest point of pronotal sclerite, in lateral view.

ML:

Mesosoma length from transition point between pronotal neck shield and anterior pronotal slope to posterior end of metapleuron.

MW:

Maximum pronotal width in dorsal view.

PEApo:

Angle formed by dorsocaudal and caudal profile lines of petiolar node, calculated with tangens functions; in M. capitatus there is a continuous shallow convexity—here the calculated angle is formed by imaginary lines (a) from upper point of node to median point of convexity, and (b) from median point of convexity to lowest point of node slope.

PigC:

Percentage of blackish pigmentation of head capsule, estimated in dorsal view and at low magnifications.

PNaa:

Maximum pronotal height, measured perpendicular to mesosomal longitudinal axis.

poGuHL:

Maximum length of gular setae with insertion points posterior to level of hind eye margin. The head was carefully adjusted, and parallax error was avoided. When the longest hair was exactly on the level of the hind eye margin, the mean was taken between this hair and the longest hair posterior to the level of the hind eye margin.

PoOc:

Postocular distance; using the cross-scaled ocular micrometer, the head was adjusted to the measuring position of CL; caudal measuring point: median occipital margin as average between peaks and valleys of microsculpture; frontal measuring point: median head at level of posterior eye margin; the average of the left and right postocular distances was calculated.

PRaa:

Maximum propodeometapleural height, measured perpendicular to mesosomal longitudinal axis.

ScBaC:

Measured in same adjustment as ScBW: distance of anterior scape base corner from anterior margin of articular neck at its median portion; see Fig. 3.

Fig. 3

Distances measured for morphometric characters ScBaC and ScBW

ScBW:

Maximum width of scape base from corner to corner, measured in dorsal view; see Fig. 3.

SL:

Maximum straight-line scape length in dorsal view, excluding the articular condyle; the values for the two scapes were averaged.

STPLd:

Maximum distance between the centre of the propodeal stigma and the posterior margin of the propodeal lobe, measured in dorso-dorso-caudo-lateral view; the specimen was carefully tilted to the position resulting in the true maximum of the distance. The region of the bulla glandulae metapleuralis was excluded.

Trochantellus:

A narrow, inconspicuous segment between femur and trochanter, usually considered as part of the femur (Allaby 1999).

The first step of the analysis consisted of sorting out the individuals of M. capitatus. This species is easily separable from M. minor and M. cf. wasmanni by qualitative (i.e. non-morphometric) assessment based on (i) the strong depression of the straight dorsal propodeal profile below the level of the high, evenly curved promesonotum; (ii) the pronounced slenderness of mesosoma and petiole; (iii) the posterior slope of the petiole node forming a very blunt angle or a shallow convexity; and (iv) a pronounced length of the scape and the legs.

The second step was sorting out those individuals of M. minor and M. cf. wasmanni that showed typical species-specific character combinations, based on qualitative assessment. Typical specimens of M. cf. wasmanni have (i) a clearly angled dorsolateral pronotum; (ii) a reduced posterior part of the psammophore; (iii) a much more protruding anterior scape-base corner; and (iv) a red-wine tinge in the reddish mesosomal pigmentation (in fresh material). Typical M. minor specimens have (i) a significantly smaller body size than M. cf. wasmanni; (ii) a fully rounded dorsolateral pronotum; (iii) a well-developed posterior part of the psammophore; (iv) a less protruding anterior scape-base corner; and (v) a strong yellowish component in the reddish mesosomal pigmentation (and never a red-wine tinge).

These two steps resulted in 60 worker specimens determined as M. capitatus, 66 as typical M. cf. wasmanni, and 48 as typical M. minor. 70 specimens of either M. cf. wasmanni or M. minor showed intermediate character combinations suggesting a possible hybrid identity, and thus remained undetermined.

Workers of Messor species show strong allometric growth and very large variation in absolute body size. As a consequence, size differences between species can disguise body-size-independent shape differences or, alternatively, expose shape differences which are just a function of body-size differences. Here, allometries were described by linear regression of character ratios against body size. After first calculating species-specific functions only from the three sets of typical specimens, the parameters of these functions were averaged to give a single over-all-species description. This overall function was then used to predict character expressions for the assumption of each individual having the same cephalic size of, in our case, 1.9 mm (for details of the procedure, see Seifert 2008).

In detail, this removal of allometric variance was performed on the 174 typical specimens of the three species and the 70 intermediate specimens of either M. cf. wasmanni or M. minor, using the following corrections (the pigmentation character PigC was not corrected for allometry, because the within-species function for M. capitatus strongly deviated from those of the other two species):

$$ \begin{array}{*{20}{c}} {{\text{PoOc}}/{\text{C}}{{\text{L}}_{{1}.{9}}} = {\text{PoOc}}/{\text{CL}}/(0.0{393} * {\text{CS}} + 0.{2779}) * 0.{3525}} \\ {{\text{LNdC}}{{\text{V}}_{{1}.{9}}} = {\text{LNdCV}}/(--{2}.{4192} * {\text{CS}} + {9}.0{742}) * {4}.{4778}} \\ {{\text{CL}}/{\text{C}}{{\text{W}}_{{1}.{9}}} = {\text{CL}}/{\text{CW}}/(--0.0{748} * {\text{CS}} + {1}.{1347}) * 0.{9926}} \\ {{\text{SL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{SL}}/{\text{CS}}/(--0.{12}0{4} * {\text{CS}} + {1}.0{374}) * 0.{8}0{86}} \\ {{\text{MW}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{MW}}/{\text{CS}}/(--0.0{444} * {\text{CS}} + 0.{6865}) * 0.{6}0{21}} \\ {{\text{Fe2L}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{Fe2L}}/{\text{CS}}/(--0.0{9}0{6} * {\text{CS}} + {1}.0{141}) * 0.{8419}} \\ {{\text{STPLd}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{STPLd}}/{\text{CS}}/\left( {--0.0{431} * {\text{CS}} + 0.{3883}} \right) * 0.{3}0{65}} \\ {{\text{PNHaa}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{PNHaa}}/{\text{CS}}/(--0.0{3}0{8} * {\text{CS}} + 0.{5}0{95}) * 0.{451}0} \\ {{\text{PRHaa}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{PrHaa}}/{\text{CS}}/(--0.0{338} * {\text{CS}} + 0.{4}0{85}) * 0.{3442}} \\ {{\text{ML}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{ML}}/{\text{CS}}/\left( {--0.{1461} * {\text{CS}} + {1}.{487}0} \right) * {1}.{2}0{94}} \\ {{\text{ScBW}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{ScBW}}/{\text{CS}}/(--{1}.{4}00 * {\text{CS}} + {12}.{549}) * {9}.{889}} \\ {{\text{ScBAC}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{ScBAC}}/{\text{CS}}/(--0.{4919} * {\text{CS}} + {5}.{389})*{4}.{455}} \\ {{\text{EL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{EL}}/{\text{CS}}/(--0.0{297} * {\text{CS}} + 0.{2465}) * 0.{19}00} \\ {{\text{PoGUHL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{PoGUHL}}/{\text{CS}}/(--0.0{138} * {\text{CS}} + 0.{2}0{4}0) * 0.{1779}} \\ {{\text{PEAp}}{{\text{o}}_{{1}.{9}}} = {\text{PEApo}}/(--{4}.{31}0 * {\text{CS}} + {156}.{9}) * {148}.{7}} \\ \end{array} $$

The allometry-corrected data were then used in a 2-class discriminant analysis with M. capitatus as one class and all other forms as the second class, using the following discriminant function (given by SPSS):

$$ \begin{gathered} {\text{D(1)}} = {5}.{186} * {\text{CS}}--{68}.{227} * {\text{PoOc}}/{\text{C}}{{\text{L}}_{{1}.{9}}}-- \hfill \\ 0.0{8}0*{\text{LNdC}}{{\text{V}}_{{1}.{9}}} + 0.0{17} * {\text{PigC}} + {22}.{166} * {\text{CL}}/{\text{C}}{{\text{W}}_{{1}.{9}}} + {4}.0{91} * {\text{SL}}/{\text{C}}{{\text{S}}_{{1}.{9}}}-- \hfill \\ {51}.{512} * {\text{MW}}/{\text{C}}{{\text{S}}_{{1}.{9}}} + {19}.{618} * {\text{Fe2L}}/{\text{C}}{{\text{S}}_{{1}.{9}}}-- \hfill \\ {23}.{356} * {\text{STPLd}}/{\text{C}}{{\text{S}}_{{1}.{9}}} + {7}.0{18} * {\text{PNHaa}}/{\text{C}}{{\text{S}}_{{1}.{9}}}-- \hfill \\ {27}.0{82} * {\text{PRHaa}}/{\text{C}}{{\text{S}}_{{1}.{9}}} + {16}.{477} * {\text{ML}}/{\text{C}}{{\text{S}}_{{1}.{9}}}-- \hfill \\ {1}.{299} * {\text{ScBW}}/{\text{C}}{{\text{S}}_{{1}.{9}}} + {1}.{784} * {\text{ScBAC}}/{\text{C}}{{\text{S}}_{{1}.{9}}} + {17}.{4}0{8} * {\text{EL}}/{\text{C}}{{\text{S}}_{{1}.{9}}}-- \hfill \\ {12}.{498} * {\text{PoGUHL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} + 0.0{5}0 * {\text{PEAp}}{{\text{o}}_{{1}.{9}}}--{6}.{9}0{6} \hfill \\ \end{gathered} $$

The classification using D(1) fully confirmed the distinctness of all 60 workers hypothesised a priori to represent M. capitatus from the collective cluster of 184 specimens of M. cf. wasmanni, M. minor and the supposed intermediates.

To give a more precise description of allometries for the collective cluster, the strongly deviating M. capitatus workers were excluded from further analysis. A second allometry correction was then calculated for the remaining 184 individuals, using only the species-specific functions for typical M. cf. wasmanni and M. minor workers, and CS = 1.9 mm. If deviating from monophasic condition, these corrections were performed in a biphasic manner. The following corrections were used (here, the pigmentation character PigC was corrected for allometry, because the within-species functions for M. cf. wasmanni and M. minor did not strongly deviate from each other):

$$ \begin{array}{*{20}{c}} {{\text{PoOc}}/{\text{C}}{{\text{L}}_{{1}.{9}}} = {\text{PoOc}}/{\text{CL}}/(0.0{425} * {\text{CS}} + 0.{2763}) * 0.{3566}---[{\text{for CS}} \leqslant {1}.{9}]} \\ {{\text{PoOc}}/{\text{C}}{{\text{L}}_{{1}.{9}}} = {\text{PoOc}}/{\text{CL}}/(0.0{327} * {\text{CS}} + 0.{2945}) * 0.{3566}---[{\text{for}}\;{\text{CS}} > {1}.{9}]} \\ {{\text{LNdC}}{{\text{V}}_{{1}.{9}}} = {\text{LNdCV}}/(--{2}.{112} * {\text{CS}} + {8}.{461}) * {4}.{623}---[{\text{for CS}} \leqslant {1}.{9}]} \\ {{\text{LNdC}}{{\text{V}}_{{1}.{9}}} = {\text{LNdCV}}/(--0.{751}*{\text{CS}} + {6}.0{5}0) * {4}.{623}---[{\text{for CS}} > {1}.{9}]} \\ {{\text{Pig}}{{\text{C}}_{{1}.{9}}} = {\text{PigC}}/(--{29}.{89}*{\text{CS}} + {1}0{8}.{2}) * {51}.{4}} \\ {{\text{CL}}/{\text{C}}{{\text{W}}_{{1}.{9}}} = {\text{CL}}/{\text{CW}}/(--0.0{657} * {\text{CS}} + {1}.{1}0{3}0) * 0.{978}0} \\ {{\text{SL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{SL}}/{\text{CS}}/(--0.{1193} * {\text{CS}} + {1}.0{1}0{2}) * 0.{7835}} \\ {{\text{MW}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{MW}}/{\text{CS}}/(--0.0{375} * {\text{CS}} + 0.{68}0{6}) * 0.{6}0{94}} \\ {{\text{Fe2L}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{Fe2L}}/{\text{CS}}/(--0.0{82}0*{\text{CS}} + 0.{98}0{7}) * 0.{825}0} \\ {{\text{STPLd}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{STPLd}}/{\text{CS}}/\left( {--0.0{416}*{\text{CS}} + 0.{3889}} \right) * 0.{3}0{98}---[{\text{for}}\;{\text{CS}} \leqslant {1}.{9}]} \\ {{\text{STPLd}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{STPLd}}/{\text{CS}}/(--0.0{378} * {\text{CS}} + 0.{3812}) * 0.{3}0{98}---[{\text{for}}\;{\text{CS}} > {1}.{9}]} \\ {{\text{PNHaa}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{PNHaa}}/{\text{CS}}/(--0.0{23}0 * {\text{CS}} + 0.{4948}) * 0.{4511}} \\ {{\text{PRHaa}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{PrHaa}}/{\text{CS}}/(--0.0{237}*{\text{CS}} + 0.{3997}) * 0.{3546}} \\ {{\text{ML}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{ML}}/{\text{CS}}/(--0.{1252}*{\text{CS}} + {1}.{4387}) * {1}.{2}00{8}} \\ {{\text{ScBW}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{ScBW}}/{\text{CS}}/(--{1}.{316} * {\text{CS}} + {12}.{3}0{2}) * {9}.{8}00} \\ {{\text{ScBAC}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{ScBAC}}/{\text{CS}}/(--0.{3934} * {\text{CS}} + {5}.0{68}) * {4}.{321}} \\ {{\text{EL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{EL}}/{\text{CS}}/(--0.0{296} * {\text{CS}} + 0.{245}0) * 0.{189}0---[{\text{for}}\;{\text{CS}} \leqslant {1}.{9}]} \\ {{\text{EL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{EL}}/{\text{CS}}/(--0.0{215} * {\text{CS}} + 0.{2298}) * 0.{189}0---[{\text{for}}\;{\text{CS}} > {1}.{9]}} \\ {{\text{PoGUHL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{PoGUHL}}/{\text{CS}}/(--0.00{43} * {\text{CS}} + 0.{2124}) * 0.{2}0{24}---[{\text{for}}\;{\text{CS}} \leqslant {1}.{9}]} \\ {{\text{PoGUHL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} = {\text{PoGUHL}}/{\text{CS}}/(--0.0{427} * {\text{CS}} + 0.{2835}) * 0.{2}0{24}---[{\text{for}}\;{\text{CS}} > {1}.{9}]} \\ {{\text{PEAp}}{{\text{o}}_{{1}.{9}}} = {\text{PEApo}}/(0.{689} * {\text{CS}} + {142}.{8}) * {144}.{1}} \\ \end{array} $$

After these corrections, the 70 workers with morphology assessed as intermediate between M. cf. wasmanni and M. minor were hypothesised to represent either mw (F1 hybrids) or mmw (hybrids backcrossed with M. minor). This grouping was based (i) on the overall visual perception of angularity of the pronotum, petiole shape, general colouration, and sculpture of the entire body, and (ii) on the values of corrected morphometric data relative to those for the specimens typical of the parental species. Subsequently, data were computed in a 2-class discriminant function (with M. minor and M. cf. wasmanni as classes) in which only the 114 specimens typical of M. minor and M. cf. wasmanni were allocated to classes a priori, while the data of the 70 intermediate specimens were run without using a-priori hypotheses; the function given by SPSS was:

$$ \begin{array}{*{20}{c}} {{\text{D}}\left( {2} \right) = 0.{155}*{\text{CS}} + {14}.{963}*{\text{PoOc}}/{\text{C}}{{\text{L}}_{{1}.{9} + }}0.{271}*{\text{LNdC}}{{\text{V}}_{{1}.{9}}} + 0.0{18}*{\text{Pig}}{{\text{C}}_{{1}.{9}}} + {32}.0{31}*} \\ {{\text{CL}}/{\text{C}}{{\text{W}}_{{1}.{9}}} + {22}.{59}0*{\text{SL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} + {33}.{355}*{\text{MW}}/{\text{C}}{{\text{S}}_{{1}.{9}}} - {35}.{615}*{\text{Fe2L}}/{\text{C}}{{\text{S}}_{{1}.{9}}} + {67}.{915}*{\text{STPLd}}/{\text{C}}{{\text{S}}_{{1}.{9}}} + } \\ {{3}.0{27}*{\text{PNHaa}}/{\text{C}}{{\text{S}}_{{1}.{9}}} - {32}.{613}*{\text{PRHaa}}/{\text{C}}{{\text{S}}_{{1}.{9}}} - {14}.{754}*{\text{ML}}/{\text{C}}{{\text{S}}_{{1}.{9}}} + 0.{27}0*{\text{ScBW}}/{\text{C}}{{\text{S}}_{{1}.{9}}} + } \\ {0.{6}00*{\text{ScBAC}}/{\text{C}}{{\text{S}}_{{1}.{9}}} - {61}.0{8}0*{\text{EL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} - {18}.{636}*{\text{PoGUHL}}/{\text{C}}{{\text{S}}_{{1}.{9}}} - 0.0{27}*{\text{PEAp}}{{\text{o}}_{{1}.{9}}} - {27}.{988}} \\ \end{array} $$

D(2) can be interpreted as describing a vector between typical M. minor and M. cf. wasmanni specimens (cf. Costedoat et al. 2005). The classification result from using D(2) for the 114 workers hypothesised to represent the parental species was in line with the a-priori hypothesis for all specimens. The remaining 70 individuals from 13 colonies, hypothesised to be hybrids or backcrosses of hybrids, were positioned fully intermediately between the parental species when sample means are considered. The position along the vector also was in line with the a-priori hypotheses (not used as classes in D(2)) for the specimens of intermediate morphology, with mmw individuals falling between mw and M. minor, and mw individuals falling between those of the parental species (Fig. 1c). This analysis did not provide a suggestion for backcrosses of hybrids with M. cf. wasmanni.

Appendix 3 Inference key for interpreting microsatellite, mtDNA and morphometric data of workers of eusocial male-haploids at the colony level

The inference key is designed for the interpretation of data on individual workers from a given colony suspected to contain genomes of more than one but of a maximum of three species, using microsatellite, mtDNA and morphometric data.

Based on the hereditary patterns established for the three data sources, the inference key classifies colonies using the following five questions. (i) What species are discernible in the colony? (ii) How many individuals are involved in the workers’ parenthood? The inference key accounts for male haploidy, in that in male haploids the maximum number of alleles per microsatellite locus consistent with a single mother mated once is three across all offspring, with at least one of the alleles occurring in all individuals (whereas for diploid organisms the corresponding value would be four, without additional constraints). If a single mtDNA haplotype has been retrieved from a colony, the presence of three alleles per microsatellite with none of these occurring in all individuals, or of more than three alleles per microsatellite, is used for the inference of either a single queen mated multiply or more than one queen per colony – no distinction is possible in these instances. The presence of more than one mtDNA haplotype is used to infer more than one queen. (iii) Is there evidence of hybridisation, i.e. gene flow across species? A colony is inferred to contain hybrids, if any of three criteria apply to at least one individual from that colony: admixed microsatellite genotype, intraindividual combination of a pure-species microsatellite genotype of one species with a mtDNA haplotype of another species, hybrid phenotype. (iv) In case of hybridisation, how far does it date back? F1 hybridisation is diagnosed against older hybridisation using the following rationales. With F1 hybrids, microsatellites, which are biparentally inherited nuclear DNA, should reveal a heterospecific 50:50 admixture, and morphology, which is nuclear encoded and thus likewise biparentally inherited, should be intermediate between species. With hybridisation older than F1, the hybrids should be more similar in microsatellites and morphology to any of the pure parental species than F1 hybrids would be. In cases of several or many backcrossings with the paternal species, both microsatellites and morphology would indicate the pure paternal species, but hybridisation older than F1 is then discernible via a mtDNA haplotype of the maternal species (instances of a high number of backcrossings with the maternal species cannot be detected via our approach, i.e. colonies to which this potentially applies are classified as pure-species colonies in step iii). (v) Likewise in case of hybridisation, is there information on its direction(s) concerning the identity of the parental species? mtDNA is used to infer the maternal species.

The precondition for using the key is that the three methods have been established to be suitable for the species concerned, i.e. that (1) each species has alleles of the microsatellite loci not shared with the other species; (2) each species has distinct, species-specific mtDNA haplotypes; (3) the hybrid worker morphologies are morphometrically diagnosable, i.e., ideally, the F1 hybrid morphology plus morphologies intermediate between the F1-hybrid morphology and the two parental species can be differentiated.

Microsatellite data must be used to assign individuals to clusters (the number of clusters being identical to the number of species suspected of being involved) using the package STRUCTURE (Pritchard et al. 2000) or similar software. The admixture of individuals to different species is then classified in steps of 25%, as follows: 1–37.5% = “25%”; 37.5–62.5% = “50%”; 62.5–99% = “75%”. Mitochondrial DNA data may be used in the key only if there is no indication of paternal leakage plus recombination (e.g. Ciborowski et al. 2007, and references therein), as these would confuse the picture from the mtDNA data and thus lead to incorrect inferences using the key. In selecting individuals for the morphometric analyses, the extremes of variation should be included to cover the whole range of occurring morphologies.

In using the inference key, the following three notes should be borne in mind. (Note 1) The key is an approximation and views the species identities of hybrid worker offspring in steps of 25%. (Note 2) The key will fail in case of repeated backcrossings with the maternal species resulting in a hybrid colony that is indiscernible from a pure-species colony never involved in hybridisation. (Note 3) If, for any colony, data from one or two disciplines are lacking, we suggest – after merging the inferences of all the colonies with complete data – to carefully make minimum inferences using plausibility arguments. This involves comparing the colony in question with colonies with complete data.

Inference key

• Step 1 View individuals separately

  • (A) In no individual is the presence of more than one species suggested by the disciplines combined (but more than one species can occur across individuals) ................................................Step 2

  • (B) In at least one individual, the presence of at least two species is suggested by the disciplines combined ......................................................Step 4

• Step 2 Across individuals

  • (A) All individuals are identical ................. consistent with one species (identity discernible from any discipline); for information on the number of individuals involved in parenthood ..............Step 3

  • (B) Individuals equal two species ............consistent with two species (identities discernible from any discipline), no gene flow between species involved; for information on the number of individuals involved in parenthood ................................. Step 3, using the data of the two species separately

  • (C) Individuals equal three species ........ consistent with three species (identities discernible from any discipline), no gene flow between species involved; for information on the number of individuals involved in parenthood .......................... Step 3, using the data of the three species separately

• Step 3 Across individuals

  • (A) mtDNA reveals a single haplotype, and microsatellites in each locus reveal a maximum of three alleles, at least one of them occurring in all individuals ......................... consistent with one father and one mother

  • (B) mtDNA reveals a single haplotype, microsatellites reveal more than three alleles in any locus, or up to three alleles, but none of them occurs in all individuals .................... consistent with more than one father and/or more than one mother

  • (C) (Only applicable if preceding step was step 5B): mtDNA reveals a single haplotype, and the alleles of all microsatellites suggest one pure species for some individuals, but the other pure species for other individuals ................. consistent with more than one father and/or more than one mother

  • (D) mtDNA reveals more than one haplotype, independently of the maximum number of alleles in each microsatellite locus.................. consistent with more than one mother, the minimum number of females involved equalling the number of haplotypes; no inference on the number of fathers

• Step 4 Across individuals

  • (A) Presence of two species is indicated ........... Step 5

  • (B) Presence of three species is indicated .......... Step 7

• Step 5 Across individuals

  • (A) In each discipline, all individuals show the same pattern, microsatellites revealing 50:50 heterospecific admixture, and worker morphology being intermediate between species ........consistent with F1 hybrids (mtDNA revealing the maternal species); for information on the number of individuals involved in parenthood ............. Step 3

  • (B) (Only applicable if the same individuals were used under all three disciplines): for some of the individuals, the presence of only one species (possibly of different identities across these individuals) is indicated by the disciplines combined; for these individuals ................................... Step 2;for the remaining individuals decide between Step 5A and 5C

  • (C) Other .......... consistent with hybridisation older than F1 (mtDNA revealing the maternal species); for information on the direction of hybridisation ............................................ Step 6; for information on the number of individuals involved in parenthood ............................. Step 3

  • Step 6 Across individuals

  • (A) mtDNA suggests one species ...........consistent with all hybridisations in the colony’s ancestry having started with females of one species (identity discernible through mtDNA)

  • (B) mtDNA suggests two species .............. consistent with the colony’s ancestry including hybridisations that started with females of two species (identities discernible through mtDNA)

  • (C) (Only applicable if preceding step was step 7B): mtDNA suggests three species .......................... consistent with the colony’s ancestry including hybridisations that started with females of three species (identities discernible through mtDNA)

• Step 7 Across individuals

  • (A) For some individuals, microsatellites and morphometrics suggest the same single species, dentical with the species suggested by mtDNA .............. consistent with one of the three species (identity discernible from any discipline) not being involved in interspecific gene flow; for information on the number of individuals involved in the colony’s parenthood concerning that species ........................ Step 3; for information on the hybrid status concerning the other two species .................................. Step 5

  • (B) Other ............. consistent with the colony’s ancestry involving hybridisations between all three species; additional interpretations are difficult, but if one of the three species is revealed solely through a very small portion in microsatellite admixture ........... Step 5, adding to any interpretation that this species was involved at some point in the colony’s ancestry, but was outcrossed subsequently; in any case: for information on the direction of hybridisation ................................................................. Step 6

Appendix 4 Results of subjecting the combined data on the single colonies to the new inference key (Appendix 3)

For details, see also the Discussion section entitled “Possible explanations for complex patterns in the colony-level inferences”.

M3::

1B→4A (M. minor, M. cf. wasmanni)→5C→6A (M. minor); →3A

M13::

1A→2A (M. cf. wasmanni)→3A

M14::

1A→2A (M. minor)→3A

M16::

1B→4A (M. minor, M. cf. wasmanni)→5C→6A (M. minor); →3A

M17::

1A→2A (M. cf. wasmanni)→3A

M18::

1A→2A (M. cf. wasmanni)→3A

M20::

1B→4A (M. minor, M. cf. wasmanni)→5A (M. cf. wasmanni)→3A

M29::

key not applicable because no data on microsatellites; minimum inference: consistent with only M. capitatus present in the colony

M40::

1B→4A (M. minor, M. cf. wasmanni)→5C→6A (M. minor); →3B

M41::

1B→4B (M. minor, M. cf. wasmanni, M. capitatus)→7A (M. capitatus)→3A; →5C→6A (M. minor); →3C

M42::

key not applicable because no data on microsatellites; minimum inferences: presence of M. capitatus and hybrids between M. minor and M. cf. wasmanni, no indication of more than one pair of parents, neither for M. capitatus nor for the hybrid of M. minor and M. cf. wasmanni

M43::

1B→4A (M. minor, M. cf. wasmanni)→5C→6B (M. minor, M. cf. wasmanni); →3D (two)

M44::

1B→4A (M. minor, M. cf. wasmanni)→5C→6B (M. minor, M. cf. wasmanni); →3D (two)

M48::

1B→4A (M. minor, M. cf. wasmanni)→5C→6B (M. minor, M. cf. wasmanni); →3D (two)

M49::

1B→4A (M. minor, M. cf. wasmanni)→5C→6B (M. minor, M. cf. wasmanni); →3D (two); this inference is based on the assumption that presence of both pure species in a colony would have resulted in their identification via the morphometric data for that colony, given that all workers remaining after molecular analysis were examined when selecting workers for morphometric analysis; here, however, all ca. 50 workers of the sample homogenously had the morphological appearance of M. cf. wasmanni, and this was confirmed by the results for the 6 individuals selected for morphometrics; hence we consider it very unlikely that the two workers suggested to be M. minor by microsatellites and mtDNA had M. minor morphology and, using a plausibility argument, consider the presence of M. minor in the molecular results to be the result of hybridisation followed by backcrossing.

M50::

1B→4B (M. minor, M. cf. wasmanni, M. capitatus)→7B, but: in no colony gene flow between M. capitatus and any other species inferred, and we assume that there was a single individual of pure M. capitatus in the colony-sample which was destroyed by DNA extraction, and therefore opt for 7A (M. capitatus)→3A; →5C→6A (M. minor); →3A

M52::

1B→4A (M. minor, M. cf. wasmanni)→5C→6A (M. cf. wasmanni); →3C

M57::

1B→4A (M. minor, M. cf. wasmanni)→5C→6B (M. minor, M. cf. wasmanni); →3D (two)

M58::

1B→4A (M. minor, M. cf. wasmanni)→5C→6A (M. cf. wasmanni); →3C

M60::

1B→4A (M. minor, M. cf. wasmanni)→5C→6A (M. minor)→3C; this inference is based on the assumption that presence of both pure species in a colony would have resulted in their identification via the morphometric data for that colony, given that all workers remaining after molecular analysis were examined when selecting workers for morphometric analysis; here, however, all ca. 50 workers of the sample homogenously had the morphological appearance of M. cf. wasmanni, and this was confirmed by the results for the 6 individuals selected for morphometrics; hence we consider it very unlikely that the worker suggested to be M. minor by microsatellites and mtDNA had M. minor morphology and, using a plausibility argument, consider the presence of M. minor in the molecular results to be the result of hybridisation followed by backcrossing.

M61::

1B→4A (M. minor, M. cf. wasmanni)→5C→6A (M. minor); →3D (two)

M62::

1B→4A (M. minor, M. cf. wasmanni)→5C→6A (M. minor); →3D (two)

M63::

1B→4B (M. minor, M. cf. wasmanni, M. capitatus)→7A (M. capitatus)→3A; →5C→6A (M. minor); →3C

MC1::

1A→2A (M. capitatus)→3A

MC2::

1A→2A (M. capitatus)→3A

MC3::

1A→2A (M. capitatus)→3A

MC11::

key not applicable because no data on microsatellites and mtDNA; minimum inference: consistent with only M. capitatus present in the colony, no indication of more than one pair of parents

MC12::

key not applicable because no data on microsatellites and mtDNA; minimum inference: consistent with only M. capitatus present in the colony, no indication of more than one pair of parents

MC13::

key not applicable because no data on microsatellites and mtDNA; minimum inference: consistent with only M. capitatus present in the colony, no indication of more than one pair of parents

Mes::

key not applicable because no data on microsatellites; minimum inference: presence of M. capitatus and hybrids between M. minor and M. cf. wasmanni, no indication of more than one pair of parents, neither for M. capitatus nor the hybrids of M. minor and M. cf. wasmanni

MM1::

1A→2A (M. minor)→3A

MM3::

1A→2A (M. minor)→3A

MW4::

1A→2A (M. minor)→3A