Genetic Population Structure and Dispersal Patterns in Formica Ants — a Review

Ann. Zool. Fennici 42: 163–177 ISSN 0003-455X Helsinki 28 June 2005 © Finnish Zoological and Botanical Publishing Board 2005

Genetic population structure and dispersal patterns in Formica ants — a review

Liselotte Sundström1, Perttu Seppä1 & Pekka Pamilo2

1) Ecology and Evolutionary Biology Unit, Department of Biological and Environmental Sciences, P.O. Box 65, FI-00014 University of Helsinki, Finland 2) Department of Biology, P.O. Box 3000, FI-90014 University of Oulu, Finland

Received 23 Nov. 2004, revised version received 27 Jan. 2005, accepted 3 Feb. 2005

Sundström, L., Seppä, P. & Pamilo, P. 2005: Genetic population structure and dispersal patterns in Formica ants — a review. — Ann. Zool. Fennici 42: 163–177.

Human impact on boreal forests has been extensive during a fairly short evolutionary time scale. Character species of boreal forests, such as Formica ants, may face loss of genetic diversity, increasing inbreeding, and decreasing gene ﬂow among extant habitat fragments owing to habitat loss and fragmentation. Here we review the genetic data on old-world boreal species of the genus Formica. In Formica ants colonies can have one or several queens (mono- and polygyny respectively) and this trait is often assumed to be linked with dispersal propensity, such that monogyne species disperse well and polygyne species disperse less well. Our analysis of the available data reveals three important aspects of the social and dispersal biology of Formica. First, the traditional division in mono- and polygyne species is too simple and we propose a population-based division into highly polygyne, weakly or moderately polygyne, and monogyne populations. Second, there is indeed an association between colony kin structure and dispersal in the predicted direction, i.e. restricted dispersal in polygyne species. However, this only holds for between-population differentiation, not within population genetic viscosity. When genetic viscosity within populations was examined

most species nevertheless showed a negative relationship between FIS and relatedness, indicating that low relatedness (many queens) is associated with reduced dispersal also locally. Only one species (F. exsecta) showed a signiﬁcant positive relationship. Finally, we predict that sex-biased dispersal may be a common trait in Formica species, although data on more species are needed to conﬁrm this.

Introduction a natural succession regime likely affect all organisms dependent on such habitats. When the The negative impact of human activities on the extent of suitable habitat patches decreases and spatial distribution and extent of boreal forests the isolation between them increases, population has been extensive during a fairly short evolu- size of species dependent on those habitats tends tionary time scale (Kilpeläinen et al. 2005 and to decline and the genetic composition of these references therein). Habitat loss and the frag- populations may change (e.g. Gaggiotti 2003). mentation of forest habitats previously under In particular, habitat loss and fragmentation 164 Sundström et al. • ANN. ZOOL. FENNICI Vol. 42 may lead to loss of genetic diversity, increasing and mires is much slower than in forests, and inbreeding, and decreasing gene flow among the recent extensive ditching of bogs has had extant habitat fragments. Species adapted to con- profound effects on ant communities in these tinuous habitats may still be locally abundant areas (Vepsäläinen et al. 2000). Human inter- and ecologically dominant, yet face problems ference is thus rapidly changing the ant com- persisting in fragmented habitats, because weak munities within the boreal region (Punttila et dispersal may prevent recolonization of the frag- al. 1994, 1996, Punttila 1996, Vepsäläinen et ments that have gone extinct (Tilman et al. 1994, al. 2000). Of the ten mound-building Formica Hanski & Gilpin 1997). species discussed in this paper (F. aquilonia, In social insects, the division into single- F. polyctena, F. rufa, F. lugubris, F. paralugu- queen (monogyne) and multi-queen (polygyne) bris, F. pratensis, F. exsecta, F. pressilabris, societies entails a range of traits that have been F. truncorum and F. sanguinea), the first six collectively coined as the “polygyny syndrome” are typical for sparse to medium dense mature (Rosengren & Pamilo 1983, Keller 1993). forest stands or forest edges, whereas the last The general view is that polygyny tends to be four are typical for ephemeral open patches that associated with short-range dispersal, whereas have arisen following disturbance (Collingwood monogyny tends to be associated with long- 1979, Seifert 1996, Czechowski et al. 2002). range dispersal through nuptial flights (Hölldo- An additional species — F. uralensis, not dis- bler & Wilson 1977, 1990, Keller 1993, Rosen- cussed here — inhabits bogs and wetland areas. gren et al. 1993, Sundström 1995a, 1995b). In its In addition to the mound-building species, the extreme, reduced dispersal in polygyny has led genus Formica also includes ground-nesting to the emergence of multi-nest colonies (poly- species, such as F. fusca, F. candida, F. cinerea, domy) where new colonies are formed near the and F. selysi and a range of less studied spe- natal colony and exchange resources and work- cies (subgenus Serviformica Seifert, 1996). Of ers with each other (Rosengren & Pamilo 1983). these, F. fusca is ubiquitous and occurs in a wide Evidence for this dichotomy in dispersal pat- range of habitats, whereas the other three are terns between monogyne and polygyne species more specialized and more patchily distributed. is based on studies of individual species (Pamilo F. cinerea typically occurs in xerotermic habi- et al. 1997, Chapuisat & Keller 1999, DeHeer et tats, F. candida (syn. picea, transkaucasica) on al. 1999, Rüppell et al. 1999, Liautard & Keller bogs and wetlands (Collingwood 1979, Seifert 2001, Seppä et al. 2005), whereas other stud- 1996, Czechowski et al. 2002), and F. selysi on ies have found curtailed dispersal and nuptial frequently inundated wetlands (Seifert 1996). flights also in monogyne species (Sundström et Whereas the mound-building Formica are usu- al. 2003). ally territorial and ecologically dominant (i.e. Red wood ants (Formica s. str.) are char- tend to exclude each other from their territories), acter species of the old world boreal zone. the Serviformica species tend to be less territo- Typically they occur in mature forests with rial and ecologically less dominant (Savolainen mixed tree stands of medium density, on bogs, & Vepsäläinen 1988, Punttila 1996). or in open areas exposed to recent disturbance Formica species have been subject to many (Collingwood 1979, Seifert 1996, Czechowski genetic studies (Crozier & Pamilo 1996, Pamilo et al. 2002). In forest areas disturbance would et al. 1997). These have shown that colony kin originally have been caused by forest fires, structure and dispersal patterns differ among but more recently forest clear-cutting and other species, which makes the genus a good candidate forms of human interference have produced for an overview of the links between kin structure similar effects. Indeed, clear-cut areas, road and the genetic structure of populations. Based sides and meadows are suitable habitat for on both observations and genetic data, For- some mound-building red wood ants adapted to mica species have traditionally been divided into ephemeral habitats (Collingwood 1979, Seifert monogyne and polygyne species. Here we use 1996, Czechowski et al. 2002). By contrast, both published and unpublished genetic data on the natural rate of successional change in bogs several Formica species and discuss (1) whether ANN. ZOOL. FENNICI Vol. 42 • Genetic population structure and dispersal patterns in Formica ants 165 the division into mono- and polygyne species is estimate gene flow separately in the two sexes warranted, or whether a different division should (Seppä et al. 2004). be used, (2) whether dispersal generally differs Wright’s (1931) island model for genetic dif- depending on colony kin structure, especially ferentiation assumes that dispersal occurs with queen number, and (3) to what extent sex-biased equal probability among all subpopulations. dispersal and limited colonization abilities may However, the degree of genetic differentiation restrict the occurrence of Formica species. may differ between pairs of subpopulations, such that some pairs are genetically more similar and others differ to a greater extent. Such differences The source of genetic structuring are expressed either as isolation by distance (IBD, Wright 1943) or genetic viscosity (GV, Genetic differences among populations arise due Hamilton 1964). Under IBD the genetic affin- to genetic drift. The rate of allele frequency ity in a system of discrete local populations change depends on the effective population size, decreases with distance, and arises when dispers- so that genetic composition changes more rapidly ing individuals can reach only their neighbouring in small populations (Wright 1931, 1951). Sev- populations. GV refers to the same phenomenon eral factors decrease the effective population size as IBD, but within populations. GV arises either as compared with the census size, and the one of when the dispersing individuals do not reach particular interest here is living in social groups. across a continuous population, or when nest The effective population size is determined by proliferation in social insects occurs by budding. the number of reproductively active individu- Thus, colony kin structure and the dispersal als, i.e. queens and males, whereas ant work- behaviour of reproductive individuals together ers are usually sterile and do not contribute to determine the current average sizes of the breed- the breeding population. In species with single- ing populations, connectivity of the populations, queen colonies, the effective population size and genetic effects of past colonization events. matches the number of nests plus the number of The sources of the data reviewed here are colony fathers. In many ants, however, colonies presented in the Appendix. We were particularly have several reproducing queens, and the effec- interested in three central measures of genetic tive population size increases as a function of population structure. First, genetic relatedness the number and relatedness of coexisting queens describes the genetic similarity of individuals (Pamilo & Crozier 1997). within a group (Pamilo 1989). Relatedness is a While mark-recapture studies can provide central parameter in the evolution of sociality, direct estimates of current dispersal, spatial because kin selection theory assumes that indi- genetic structuring of populations studied by viduals involved in an altruistic interaction are using nuclear genetic markers (e.g. allozymes related (Hamilton 1964). Here we used related- or DNA microsatellites) can provide a picture ness among worker nestmates to describe colony of the typical dispersal ranges of individuals, as kin structure. Second, the fixation indices FIS well as footprints of past colonization events. and FST describe how genetic variation is distrib- The observed differentiation between popula- uted in a subdivided population (Wright 1931, tions is then a balance between drift and gene 1943, 1951, Weir & Cockerham 1984). FIS is flow. Alternatively, the observed structuring can the inbreeding coefficient, which measures the be a result of a founder effect rather than ongo- departure from random mating within subpopu- ing but restricted gene flow, i.e. provide foot- lations, and can take values from strongly nega- prints of historical colonization and other demo- tive (total outbreeding) and 1 (total inbreeding). graphic events (Wright 1969, Hedrick 1999). It results from the sampled population being Further details of gene flow can be obtained by genetically further subdivided (Wahlund effect, using maternally inherited mitochondrial mark- Wahlund 1928), or more rarely, from active ers. They directly reveal the amount of female mate choice favouring relatives. In our use here gene flow (Ennos 1994), and the parallel use of FIS represents the combination of both genetic nuclear and mitochondrial markers allows us to viscosity and mating between relatives as we 166 Sundström et al. • ANN. ZOOL. FENNICI Vol. 42 cannot, based on the available data, discriminate (F > 0), either due to disparity between the local between the two. FST is a measure for genetic dif- sampling site and the actual breeding popula- ferentiation among subpopulations, and can take tion (Wahlund effect, Wahlund 1928) or mating values from 0 to 1. among relatives. In both cases the relatedness In addition we extracted information on the estimates are boosted as compared with those spatial scale of sampling to allow adjustments for the ideal random mating population and the for different sampling scales. We estimated the number of breeders inferred from the related- sampling areas both within and across sampling ness values will be underestimated. Therefore, sites (local and global scales, respectively), and we adjusted all relatedness estimates used in the included the relevant scale as a covariate in all analyses for inbreeding (Pamilo 1985). analyses. In some studies the approximate area As noted previously by Pamilo et al. (1997), was reported and in some cases the sampling the average relatedness in Formica varies con- sites were checked from maps and the areas cal- siderably both among and within species, and culated based on the distribution of the sampled in none of the species indicates consistent mono- nests. Finally, sometimes the sampling was done gyny (Table 1). In F. aquilonia, F. polyctena and along transects and here we assumed transects to F. paralugubris relatedness is invariably low be 20 m wide, and calculated the area based on it. indicating the obligate presence of many repro- Thus, the estimates of the sampling area are not ductively active queens. Indeed, excavations and precise, but should correctly reflect the orders of calculations based on genetic data indicate that magnitude in sampling areas. The nuclear data the queen number frequently rises to several comprise studies made by using both allozymes hundreds (Rosengren et al. 1993 and references and DNA microsatellites as genetic markers therein). Nevertheless, relatedness only rarely (Appendix). As the estimates based on these two equals zero (but see Elias et al. 2005), suggest- classes of markers may not be entirely compara- ing some degree of relatedness structure despite ble (Hedrick 1999) we included the marker type high numbers of queens. Interestingly, F. cinerea as a covariate in all analyses. Whenever a vari- (Goropashnaya et al. 2001), F. lugubris (Gyl- able in the analyses did not conform to normality lenstrand & Seppä 2003), F. exsecta (Pamilo we used logarithm-transformed values (ln x + 1), & Rosengren 1984, Seppä et al. 2004) and F. which led to fulfillment of the requirements of truncorum (Sundström 1993) are socially poly- normality in all cases. morphic, so that the average population-specific relatedness can vary from close to zero to 0.75, the expected value under monogyny/monandry. Patterns of intra-colony Finally, in several species the average related- relatedness ness across populations is intermediate. This indicates either the presence of a few queens Relatedness is a relative measure of similarity and/or multiple mating by queens, or exten- among group members. In principle, the correct sive within-population variation in relatedness reference population in the estimation procedure so that some nests within a population have a is the breeding population (Pamilo 1989, Queller single queen, while others are highly polygyne. & Goodnight 1989), but this is usually difficult In F. selysi colonies within the same population to assess in the field, and for practical reasons indeed are either monogyne or highly polygyne the local sampling site is considered as the actual (Chapuisat et al. 2004). breeding population. In ants, colonies are family Monogyny and polygyny refer to the state units, and relatedness among the brood directly of a colony, and the above examples show that reflects the breeding structure in the colonies i.e. the division cannot always be extended easily the number of reproducing queens and males and to concern populations or species. Therefore the relatedness among them (Ross 1993), as well a biologically more relevant division seems to as the pattern of reproductive partitioning among be either “obligately” vs. “facultatively poly- these. However, the inbreeding coefficient often gyne species”, with the latter group including indicates some degree of non-random mating also socially polymorphic species, or “highly ANN. ZOOL. FENNICI Vol. 42 • Genetic population structure and dispersal patterns in Formica ants 167 polygyne”, “weakly/moderately polygyne” and between species. When the entire data set was “monogyne” species/populations. The former considered the average estimate of relatedness classification goes along the species boundaries, was r = 0.28 ± 0.26 (mean ± SD) and r = 0.22 whereas the latter emphasizes the prevailing ± 0.24, for allozyme and DNA-microsatellite status of each population and concedes the high data, respectively; T = 1.26, d.f. = 94, p = 0.21. degree of flexibility in queen numbers in the When only those species were considered for genus. In the highly polygyne group, we include which both allozyme and microsatellite data the obligately polygyne species F. aquilonia, F. were available the corresponding values were: polyctena and F. paralugubris, as well as the 0.22 ± 0.27, 0.23 ± 0.25, T = –0.14, d.f. = 65, p = polygyne populations (r < 0.25) of the socially 0.89. The mean pairwise difference between the polymorphic species F. exsecta, F. truncorum, markers within species was 0.026 ± 0.035, T = F. lugubris, F. selysi and F. cinerea. The weakly/ 0.35, d.f. = 6, p = 0.71. moderately polygyne group encompasses species with higher intracolony relatedness (r > 0.25), excluding the purely monogyne popula- Patterns of spatial genetic tions of socially polymorphic species (r > 0.50) structuring (Table 1). Given the average degree of multiple mating by queens a relatedness of 0.50 can still The general consensus has been that monog- be compatible with monogyny (Pamilo 1993, yny (high intra-colony relatedness) is associated Sundström 1993, Boomsma & Ratnieks 1996, with extensive dispersal and outbreeding, and Hannonen et al. 2004). In our analysis below we polygyny (low intra-colony relatedness) is asso- have treated each population as an independent ciated with restricted dispersal and inbreeding sample, and classified them as highly polygyne, (Hölldobler & Wilson 1990). Restricted disper- weakly/moderately polygyne or monogyne. Our sal, especially by females, and observations of justification for this is that queen number is such local matings on nest mounds indeed suggest that a highly variable trait that different dispersal inbreeding could be more common in polygyne regimes may prevail within species as well as species (Hölldobler & Wilson 1990), although a

Table 1. Average inbreeding-corrected relatedness (r ), inbreeding (FIS), population differentiation (FST), isolation by distance (IBD) and genetic viscosity (GV) per species. The sample sizes (n) refer to the number of populations

(FIS and r ) or sets comprising at least two populations (FST) that were used; standard deviations were calculated over populations (FIS and r ) or sets of populations (FST); for IBD and GV, the ﬁrst value indicates the number of populations in which IBD or GV was found (either in nuclear and/or mitochondrial markers), and the second number indicates the number of populations studied.

r F F IBD GV IS ST mean ± SD n min/max mean ± SD n min/max mean ± SD n min/max

F. aquilonia 0.02 ± 0.08 19 –0.12/0.15 0.03 ± 0.05 19 –0.08/0.17 0.17 ± 0.07 3 0.09/0.24 1/1 0/1 F. candida 0.23 1 – 0.07 ± 0.10 1 – – – – – 0/1 F. cinerea 0.24 ± 0.31 15 –0.10/0.81 0.04 ± 0.07 15 –0.05/0.24 0.06 ± 0.03 5 0.03/0.10 0/1 1/8 F. exsecta 0.34 ± 0.25 21 –0.01/0.73 0.03 ± 0.10 21 –0.12/0.19 0.09 ± 0.05 4 0.05/0.15 0/4 6/10 F. fusca 0.48 ± 0.17 6 0.25/0.67 0.05 ± 0.08 6 –0.07/0.17 0.09 1 – 0/1 – F. lugubris 0.21 ± 0.21 6 –0.02/0.54 0.03 ± 0.08 11 –0.11/0.17 0.07 ± 0.06 4 0.02/0.16 – 1/1 F. paralugubris –0.03 ± 0.03 2 –0.05/–0.01 0.13 ± 0.04 2 0.10/0.16 – – – – 1/1 F. polyctena 0.11 ± 0.05 6 0.01/0.16 0.17 ± 0.06 6 0.08/0.26 0.07 ± 0.06 2 0.02/0.11 – 0/1 F. pratensis 0.44 ± 0.26 4 0.15/0.66 –0.04 1 – – – 1/1 F. pressilabris 0.18 ± 0.16 2 0.07/0.29 –0.03 ± 0.03 2 –0.05/–0.01 0.09 ± 0.00 1 – – – F. rufa 0.45 ± 0.12 9 0.27/0.59 0.09 ± 0.11 9 –0.06/0.27 0.04 ± 0.02 3 0.02/0.05 – 1*/2 F. sanguinea 0.44 ± 0.07 6 0.31/0.54 0.04 ± 0.10 6 –0.08/0.18 0.03 1 – – 2/3 F. selysi 0.40 ± 0.44 2 0.09/0.71 0.04 ± 0.01 2 0.03/0.05 0.00 ± 0.00 2 0.00/0.01 – 0/1 F. truncorum 0.29 ± 0.27 11 –0.07/0.67 0.02 ± 0.10 11 –0.12/0.18 0.09 ± 0.08 4 0.03/0.19 – 0/1

* viscosity found only in resident colony queens. 168 Sundström et al. • ANN. ZOOL. FENNICI Vol. 42 high queen number in a nest means that intranidal We repeated this analysis but entered average mating does not necessarily occur between close relatedness instead of population type and added relatives. Thus we would expect the estimate of the interaction relatedness ¥ species in our GLM inbreeding (FIS), reflecting either within-popula- model. As expected, based on the analysis above, tion viscosity or mating among relatives, to be relatedness had no significant effect on FIS, but higher under polygyny than under monogyny. both the effect of species and scale as well as the When the populations were classified as highly interaction between relatedness and species were polygyne, weakly polygyne and monogyne, treat- significant (GLM with stepwise backward elimi- ing each population as an independent sample we nation, species: F12,85 = 2.2, p = 0.02, scale: F1,85 ¥ found no significant effects of population type, = 6.8, p = 0.01, species relatedness: F13,85 = 2.6, marker type or species on inbreeding (FIS) (Table p = 0.004, the effects of marker and relatedness 2). Only the scale of sampling had a significant were eliminated under threshold criteria of p > effect, with the estimate of FIS increasing with 0.15) (Fig. 1a). This indicates that the degree of sampling area (Table 2). This lack of difference within-population viscosity as estimated by FIS is largely due to the divergent pattern found in does not change in concert with queen number F. exsecta, where high estimates of inbreeding across species. Instead the species differ in their were obtained in several monogyne populations response such that the negative relationship (Sundström et al. 2003, Seppä et al. 2004). When holds in some species, but not in others (Fig. this species was omitted from the analysis the 2). Although the limited number of replicates in difference between population types was sig- some species precludes more robust conclusions, nificant (GLM: F2,74 = 5.2, p = 0.008, species: several species show significant negative trends

F12, 74 = 1.01, p = 0.45, marker type: F1,74 = 0.17, and one species (F. exsecta) shows a significant p = 0.68, scale: F1,74 = 2.2, p = 0.14). In this data positive trend (Fig. 2). However, the pattern may set, the average degree of inbreeding decreased change when more data becomes available. to 0.003 ± 0.07 for monogyne populations and Overall the average values of FIS tended to be remained unchanged for weakly and highly poly- positive, but in many individual cases not statis- gyne populations. The very high estimates of FIS tically different from zero (Table 1 and Appen- obtained in monogyne populations of F. exsecta dix). Some rather high values obtained with may be due to the small number of colonies allozymes were not significant although similar sampled (3–5 colonies in most cases; Seppä et al. and much lower values obtained with microsatel- 2004). Nevertheless, inbreeding was also high in lites were significant (Appendix). This highlights a monogyne population where nearly 70 colonies the problem of power in analyses based on few had been sampled (Sundström et al. 2003). loci. We feel, however, justified in using all the

Table 2. Average degree of inbreeding (FIS), population differentiation (FST) and nest mate relatedness (r ) in highly polygyne, weakly polygyne and monogyne species/populations. Standard deviations calculated over the different studies. The lower half gives the results of a GLM with each of the three parameters as dependent variables.

r F F IS ST mean ± SD n min/max mean ± SD n min/max mean ± SD n min/max

Highly polygyne 0.05 ± 0.8 60 –0.12/0.23 0.05 ± 0.08 60 –0.12/0.26 0.10 ± 0.07 15 0.01/0.24 Weakly polygyne 0.39 ± 0.08 27 0.25/0.48 0.06 ± 0.10 26 –0.12/0.27 0.06 ± 0.03 7 0.02/0.09 Monogyne 0.62 ± 0.07 26 0.52/0.81 0.02 ± 0.08 26 –0.12/0.19 0.03 ± 0.02 8 0.00/0.05

GLM F d.f. p F d.f. p F d.f. p

Population type 329.6 2,94 < 0.0001 0.96 2,94 0.38 4.15 2,16 0.03 Sampling scale 0.30 1,94 0.58 5.64 1,94 0.02 0.06 10,16 0.37 Species 1.75 13,94 0.06 0.80 13,94 0.66 1.46 1,16 0.24 Marker 0.47 1,94 0.49 0.08 1,94 0.78 0.11 1,16 0.74 ANN. ZOOL. FENNICI Vol. 42 • Genetic population structure and dispersal patterns in Formica ants 169

0.3 0.3 a b

0.2

0.2 0.1 IS ST F F 0 0.1

–0.1

–0.2 0 –0.2 0 0.2 0.4 0.6 0.8 1 0 0.25 0.5 0.75 1 Relatedness (r) Relatedness (r)

Fig. 1. The relationship between relatedness and FIS (a) and FST (b) across all populations.

F. aquilonia F. cinerea F. exsecta 0.2 0.3 0.3 r = –0.54, n =19, p = 0.01 r = -0.45, n =16, p = 0.08 r = 0.47, n = 21, p = 0.03 0.2 0.2 0.1 0.1

IS 0.1

F 0 0 0 –0.1

–0.1 –0.1 –0.2 –0.2 0 0.2 –0.2 0 0.2 0.4 0.6 0.8 1 –0.2 0 0.2 0.4 0.6 0.8 1

F. fusca F. lugubris F. polyctena 0.2 0.2 0.3 r = -0.71, n = 6, p = 0.11 r = –0.69, n = 6, p = 0.13 0.1 0.1 0.2 IS

F 0 0 0.1 –0.1 –0.1 r = 0.05, n =11, p = 0.88 0 0.2 0.4 0.6 0.8 1 –0.2 0 –0.2 0 0.2 0.4 0.6 0.8 1 0 0.2 F. rufa F. sanguinea F. truncorum 0.3 0.2 0.2 r = –0.70, n = 9, 0.2 p = 0.03 0.1 0.1 IS

F 0.1 0 0 0 –0.1 r = –14, n = 6, r = –0.59, n = 11, p = 0.79 p = 0.05 –0.1 –0.1 –0.2 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 –0.2 0 0.2 0.4 0.6 0.8 1 Relatedness (r) Relatedness (r) Relatedness (r)

Fig. 2. The relationship between FIS and relatedness for all species separately. available values as they represent unbiased esti- was entered in all analyses it is unlikely that mates of the actual degree of within population the marker type has influenced the conclusions. viscosity, albeit with wide confidence intervals. Nevertheless, in all highly polygyne species (F. If no inbreeding were present, the average across aquilonia, F. paralugubris and F. polyctena) and 96 samples should indeed be zero, which it is not several facultatively polygyne species (F. rufa, (mean = 0.05, SD = 0.08, t = 5.8, d.f. = 95, p < F. lugubris, F. exsecta, F. fusca and F. cinerea) 0.001). This indicates that some degree of genetic a significant degree of inbreeding was found at viscosity is present in Formica ants regardless of least in some populations (Chapuisat & Keller the number of queens, although viscosity may 1999, Goropashnaya et al. 2001, Ranta 2002, be absent or not detectable in individual popu- Sundström et al. 2003, Hannonen et al. 2004, lations. As the values obtained with allozymes Gyllenstrand et al. 2005, Pamilo et al. 2005). were not consistently higher than those obtained The facultatively polygyne group also included with microsatellites and because the marker type monogyne populations. In the case of F. rufa 170 Sundström et al. • ANN. ZOOL. FENNICI Vol. 42 the spatial scale of sampling was relatively large genetic viscosity in five of six cases (F. exsecta, (200 ha), but in the other monogyne populations four populations (Liautard & Keller 2001, Seppä the scale of sampling was usually less than 20 ha. et al. 2004); F. lugubris, one population (Gyl- The high degree of inbreeding in the monogyne lenstrand & Seppä 2003)). All these populations population of F. exsecta at a very limited spatial were highly polygyne, and the range of the popu- scale (15 ha) (Sundström et al. 2003) shows that lations was similar to that in the nuclear studies monogyny is not necessarily always associated (< 500 m). This indicates that reduced female with extensive dispersal. Interestingly, F. exsecta dispersal at local scales is common at least in is the only species with dimorphic males, with polygyne species and highlights the need for male size apparently corresponding to their dis- corresponding studies in weakly or moderately persal propensity (Fortelius et al. 1987). polygyne species. Apart from mating between relatives, the As with within-population differentiation we coefficient of inbreeding (FIS) may take positive may also expect greater between-population dif- values either because a population comprises ferentiation in poly- than in monogyne popula- several reproductively isolated sets of subpopu- tions, if the dispersal propensities of the two lations or because colonies at greater distance types differ. Indeed, the between-population differ more owing to colony reproduction by bud- differentiation estimate of FST was significantly ding (GV). Genetic viscosity in nuclear markers lower in monogyne than in highly polygyne sets has been studied in twelve species in a total of 28 of populations (Table 2, Tukey post-hoc tests: p = populations. The distances between the nests in 0.01, all other p values greater than 0.18). Simi- these studies ranged from a few meters to more larly, FST decreased significantly with increas- than two kilometers. In twelve populations (43%), ing relatedness (GLM with stepwise backward a significant degree of genetic viscosity (GV) was elimination, relatedness: F1,29 = 8.4, p = 0.007; the detected. This is most likely an underestimate, remaining factors, species, local scale or marker because GV cannot be detected if the sampling type were all eliminated with the threshold crite- area is totally homogenous, e.g. if it comprises a rion for exclusion set at p > 0.15; Fig. 1b). Owing single polydomous colony. This is entirely pos- to the small number of population sets per species sible in some of the smaller sampling areas. In an analysis with the interaction species ¥ related- six cases the significant outcomes encompassed ness was not considered appropriate. This shows highly polygyne populations, whereas in six cases that sets of populations with on average low the populations were monogyne or weakly poly- relatedness also show higher between-population gyne. Of the non-significant outcomes five cases genetic differentiation, and the conclusion is that encompassed monogyne populations and eleven high queen numbers translate into reduced long- cases polygyne populations (Fisher’s exact test: range dispersal. This is even more obvious when p = 0.31). In cases where several populations remembering the connection between genetic dif- had been tested (F. cinerea, F. exsecta, F. rufa, F. ferentiation and effective population size: high sanguinea), both significant and non-significant queen numbers translate into greater effective outcomes were found (Table 1). The smallest population sizes, which should slow down the range at which isolation by distance was found effects of genetic drift producing genetic differ- was 100 m (F. exsecta) and five of the significant entiation. A more parsimonious scenario is, how- outcomes were found at scales of 500 m or less ever, that most highly polygyne populations are (F. cinerea, F. lugubris, and F. pratensis once, not in drift-migration balance, but are founded F. exsecta twice). Inbreeding, relatedness, and through a bottleneck by only one or a few immi- scale had no significant effect on the presence/ grants. This results in random representation of absence of genetic viscosity (logistic regression: only a few alleles in each locus. all p values greater than 0.59). Nevertheless, the Between-population differentiation may results show that genetic viscosity can be found follow either the infinite island model, or the in more than one third of the populations, also stepping-stone model (Wright 1931, 1951). In the at rather limited scales (a few hundred meters). latter case more distant populations are predicted Interestingly, mitochondrial markers indicated to differ more owing to isolation by distance. ANN. ZOOL. FENNICI Vol. 42 • Genetic population structure and dispersal patterns in Formica ants 171

Isolation by distance, decreasing genetic affinity as well as between sympatric monogyne and of local populations with distance, has been stud- polygyne populations. The degree of isolation ied only in seven populations of four species, obtained with the maternally inherited marker and only one of them shows IBD (Table 1). In F. was substantially higher than that obtained for aquilonia, F. exsecta and F. fusca, the distances the nuclear marker, and only the degree of dif- between local populations was between 0.2 and ferentiation among monogyne populations of F. 50 km (Sundström et al. 2003, Helanterä 2004, exsecta was not significant for both classes of Seppä et al. 2004, Pamilo et al. 2005), but in F. markers (Table 3; Seppä et al. 2004). This indi- cinerea IBD was assessed in an area covering cates that males disperse at a much higher rate most of northern Europe (max distances > 1000 than females. However, an additional study on km, Goropashnaya et al. 2001). The studies also a monogyne population of F. exsecta found sex- include one where genetic population structure biased gene flow with reduced female dispersal was studied with mitochondrial markers (Lia- at very small scales (300 m) (Sundström et al. utard & Keller 2001). 2003). This study was based on nuclear mark- The above estimates of genetic differentiation ers only, but detection of sex-biased gene flow are based on nuclear markers, but these usually was possible because post-dispersal genotypes of do not reveal the relative contributions of the both parents in each nest could be inferred from two sexes to the observed genetic differentiation worker genotypes. In a similar study on F. rufa, pattern (but see Sundström et al. 2003). Indeed, genetic viscosity was apparent in colony queens, foundation of new populations may be severely but not in colony fathers (Ranta 2002). The effect constrained owing to restricted female dispersal, was, however, weaker than in F. exsecta so no yet gene flow mediated exclusively by males significant viscosity was detected in the offspring may completely homogenize allele frequencies workers. In F. truncorum two adjacent mono- at nuclear markers, as for instance in polygyne gyne and polygyne populations did not share populations of Solenopsis invicta (e.g. Ross & any haplotypes in mitochondrial markers (N. Shoemaker 1997). In the extreme, males may Gyllenstrand et al. unpubl. data). This indicates both mate in their natal colony and later disperse a complete absence of female gene flow between thereby effectively contributing to the homogeni- the two population types, despite distances as zation of allele frequencies between populations. short as a few hundred meters. Nuclear mark- To date both nuclear and mitochondrial markers ers, on the other hand, showed moderate levels have been used to compare male and female gene of differentiation between the population types, flow in three Formica species. The most exten- which must be due to male dispersal between the sively studied species is F. exsecta. In one case two populations. Finally, in a highly polygynous the data allow comparisons of gene flow both F. lugubris population female gene flow among within mono- and polygyne sets of populations subpopulations was also negligible, but nuclear

Table 3. Local differentiation when both nuclear (FST) and mitochondrial (FST) differentiation has been estimated from the same populations. #pops is the number of subpopulations in the study; type is the type of comparison in each case (M populations; P populations or all mixed); scale is the approximate distance between subpopulations in km. Asterisks mark estimates significantly greater than zero. The ratio FST/FST = 3 indicates equal gene flow among populations by both sexes (see Seppä et al. 2004), and ratios below and above three indicate an excess of male and female gene flow, respectively.

Species/population #pops Type Scale km FST FST FST/FST

F. exsecta /Uppsala 4 M-M 0.2–9 0.00 0.06 – F. exsecta /Uppsala 6 P-P 0.2–9 0.10 0.46* 4.7 F. exsecta /Uppsala 10 M-P 0.2–9 0.06* 0.27* 4.3 F. exsecta /Åland 11 All 2–20 0.09* 0.13* 1.5 F. lugubris 5 P-P 1–3 0.03 0.53 15.6 F. truncorum 2 M-P min 0.3 0.12 1.0 8.3 172 Sundström et al. • ANN. ZOOL. FENNICI Vol. 42 genetic structure was shallow due to extensive is associated with dispersal propensity. Although gene ﬂow (Gyllenstrand & Seppä 2003). a wealth of studies address genetic population In F. exsecta, the use of mitochondrial mark- structure in conjunction with colony kin struc- ers also revealed cases where multiple matrilines ture none have to our knowledge tested this basic coexist in the same polygyne nest. In two differ- prediction across species. Indeed our analyses ent studies, eighteen (Liautard & Keller 2001) verify that this assumption is valid and that poly- and sixteen (Seppä et al. 2004) percent of polygyne populations or species also tend to have gyne nests contained more than one mtDNA hap- reduced long-range (between-population) disper- lotype, which means that immigrating females sal. However, this association did not hold for have been accepted into existing polygyne nests short-range (within-population) dispersal across at some point. Polygyne Formica colonies do species. Instead the response varied among spe- readily accept alien queens (Fortelius et al. 1993, cies with most of them showing a negative

Brown et al. 2003), but whether these act as association between FIS and relatedness. This intraspecific parasites or become subordinate genetic viscosity is most likely due to colony reproducers that only specialise in worker pro- reproduction by budding. Interestingly, one spe- duction is not known. cies (F. exsecta) showed a significant positive Finally, species adapted to ephemeral habi- trend which is mainly due to a high frequency tats are likely to be better dispersers than spe- of inbred monogyne populations (Sundström cies adapted to continuous habitats. This should et al. 2003, Seppä et al. 2004). In this spe- lead to lower degrees of between-population cies sex-biased dispersal is pronounced, with differentiation. However, colonization is often strong female philopatry in polygyne popula- also associated with founder effects, which may tions (Seppä et al. 2004). Monogyne populations counteract homogenization of allele frequencies are often fairly small and it seems likely that the across populations unless new immigrants con- high inbreeding coefficients in these populations tinuously arrive to the area. We compared relat- indeed are due to non-random mating. The fact edness, inbreeding and population subdivision that this species also has size-dimorphic males between species associated with ephemeral habi- with different dispersal propensities raises new tats and species associated with mature boreal questions regarding its reproductive biology. forests. Although the FIS was lower for species Our analysis also shows that the traditional found in ephemeral habitats than forest dwellers, division into monogyne and polygyne species the difference between the two types of species is not well supported. Instead, populations of was not significant (mean ± SD: 0.04 ± 0.08 many species show the entire range of variation and 0.06 ± 0.09, respectively; t = 1.63, p = 0.11, in dispersal and colony kin structure. None of d.f. = 94). The average estimates of FST were of the species is exclusively monogyne, although similar magnitude in both categories: 0.07 ± 0.05 several species have largely monogyne popula- and 0.08 ± 0.07, respectively; and did not differ tions. The trait (queen number) clearly is phy- significantly (t = 0.4, d.f. = 30, p = 0.69). How- logenetically very flexible within this genus (for ever, the species associated with more ephemeral a more formal analysis see Helanterä 2004). habitats have significantly higher average relat- Interestingly, only one species has so far been edness than the forest-dwelling ones (0.33 ± 0.26 shown to have unicolonial populations with zero and 0.16 ± 0.21; t = 3.54, d.f. = 94, p = 0.0006). within-colony relatedness (F. truncorum; Elias Taken together, the type of habitat (fragmented et al. 2005), although several species have high vs. continuous) to which a species is adapted numbers of queens (Rosengren et al 1993). Also does not strongly affect population parameters. in the case of F. truncorum unicoloniality is likely to be a matter of scale of sampling. Most species show some degree of popu- Conclusions lation viscosity or at least reduced dispersal also at very local scales. Given that polygyne Here we have quantified and statistically tested species regularly have large effective popula- the general assumption that colony kin structure tion sizes the effects of genetic drift should be ANN. ZOOL. FENNICI Vol. 42 • Genetic population structure and dispersal patterns in Formica ants 173 less pronounced and therefore this result indi- viscosity in many populations of Formica ants cates strongly reduced gene flow. However, the and this degree of viscosity can vary with kin alternative explanation that most populations of structure both across populations within species highly polygyne species result from single colo- and across species. nization events seems much more likely. Hence, these species may face severe colonization problems given the current rate of habitat fragmen- Acknowledgements tation. This is further strengthened by the fact that many of the species studied so far appear We would like to dedicate this review to our colleague Rainer to have some degree of sex-biased dispersal Rosengren, who always supported our careers in the world of boreal wood ants with kindness and inspiration. Sadly, he with females being more philopatric than males. is not among us any more. Several colleagues shared their as Thus, colonization of new habitable areas may in yet unpublished data and manuscripts, which considerably fact be even more difficult than indicated by the complemented the contents of this review. They deserve our observed overall degree of population viscosity sincere thanks: Marianne Elias, Hannaleena Mäki-Petäys, in these ants. Wee Tek Tay. Financial support for our work has been provided by the Academy of Finland (decision notification Finally, the species adapted to more ephem- numbers 206505/Sundström and 77311/Pamilo) as well as eral habitats, such as F. exsecta, F. truncorum, the EU-IHP network INSECTS (contract number HPRN-CT- F. sanguinea, and the Serviformica species do 2000-00052). not show significantly lower estimates of FIS or

FST than the rest of the species associated with continuous forest habitat. Nevertheless, also References these species can show extreme within-population genetic viscosity as evidenced by F. exsecta. Bernasconi, C., Maeder, A., Cherix, D. & Pamilo, P. 2005: Diversity and genetic structure of the wood ant Formica Given that the ephemeral habitats also are highly lugubris in unmanaged forests. — Ann. Zool. Fennici fragmented founder effects in conjunction with 42: 189–199. low immigration rates and a reasonable persist- Beye, M., Neumann, P., Chapuisat, M., Pamilo, P. & Moritz, ence time of the habitat, could in fact also turn R. F. A. 1998: Nestmate recognition and the genetic the balance in the opposite direction and lead to relatedness of nests in the ant Formica pratensis. — Behav. Ecol. Sociobiol. 43: 67–72. strong population subdivision in these species. Boomsma, J. J. & Ratnieks, F. L. W. 1996: Paternity in euso- When habitat persistence is very limited colo- cial Hymenoptera. — Phil. Trans. Proc. Roy. Soc. B 351: nization rates necessarily must be high and no 947–975. such differentiation is expected. One additional Chapuisat, M. & Keller, L. 1999: Extended family structure reason for the lack of a stronger substructuring in in the ant Formica paralugubris: the role of breeding system. — Behav. Ecol. Sociobiol. 46: 405–412. these species may be rapid population turnover, Chapuisat, M., Bocherens, S. & Rosset, H. 2004: Variable where populations frequently go extinct and new queen number in ant colonies: No impact on queen turn- habitats are quickly colonized by immigrants. over, inbreeding, and population genetic differentiation Conversely, large effective population sizes in in the ant Formica selysi. — Evolution 58: 1064–1072. conjunction with limited time span since habitat Chapuisat, M., Goudet, J. & Keller, L. 1997: Microsatellites reveal high population viscosity and limited dispersal fragmentation may explain the as yet moderate in the ant Formica paralugubris. — Evolution 51: degree of structuring in forest dwellers. Alterna- 475–482. tively, some of these species are reasonably good Collingwood, C. A. 1979: The Formicidae (Hymenoptera) colonizers, whereas others indeed have very lim- of Fennoscandia and Denmark. — Fauna Ent. Scandi- navica 8: 9–174. ited colonization abilities. This is consistent with Crozier, R. H. & Pamilo, P. 1996: Evolution of social insect our result that at least the FST values differed sig- colonies — sex allocation and kin selection. — Oxford niﬁcantly among species, with the highly poly- Univ. Press, Oxford. gyne species (low relatedness) having the highest Czechowski, W., Radchenko, A. & Chechowska, W. 2002: degree of population subdivision. Within-popu- The ants (Hymenoptera, Formicidae) of Poland. — Museum and Institute of Zoology PAS, Warsawa. lation genetic viscosity was however of the same DeHeer, C. J., Goodisman, M. A. D. & Ross, K. G. 1999: magnitude across all species. Taken together our Queen dispersal strategies in the multiple-queen form analysis suggests some degree of population of the ﬁre ant Solenopsis invicta. — Am. Nat. 153: 174 Sundström et al. • ANN. ZOOL. FENNICI Vol. 42

660–675. Mäki-Petäys, H., Zakharov, A., Viljakainen, L., Corander, J. Elias, M., Rosengren, R. & Sundström, L. 2005: Seasonal & Pamilo, P. 2005: Genetic changes associated to declin- polydomy and unicoloniality in a polygynous population ing populations of Formica ants in fragmented forest of the red wood ant Formica truncorum. — Behav. Ecol. landscape. — Mol. Ecol. 14: 733–742. Sociobiol. 57: 339–349. Olsson, C. 1999: Sukuyksilötuotanto ja sukulaisuusasymmet- Ennos, R. A. 1994: Estimating the relative rates of pollen and ria mustamuurahaisella (Formica fusca, Hymenoptera, seed migration among ant populations. — Heredity 72: Formicidae). — M.Sc. thesis, University of Helsinki, 250–259. Finland. Fortelius, W., Pamilo, P., Rosengren, R. & Sundström, L. Pamilo, P. 1981: Genetic organization of Formica sanguinea 1987: Male size dimorphism and alternative reproduc- populations. — Behav. Ecol. Sociobiol. 9: 45–50. tive tactics in Formica exsecta ants (Hymenoptera, For- Pamilo, P. 1982: Genetic population structure in polygynous micidae). — Ann. Zool. Fennici 24: 45–54. Formica ants. — Heredity 48: 95–106. Gaggiotti, O. E. 2003: Genetic threats to population persist- Pamilo, P. 1985: Effect of inbreeding on genetic relatedness. ence. — Ann. Zool. Fennici 40: 155–168. — Hereditas 103: 195–200. Goropashnaya, A. V., Seppä, P. & Pamilo, P. 2001: Social Pamilo, P. 1989: Estimating relatedness in social groups. and genetic characteristics of geographically isolated — Trends in Ecology and Evolution 4: 353–355. populations of the ant Formica cinerea. — Mol. Ecol. Pamilo, P. 1993: Polyandry and allele frequency differ- 10: 2807–2818. ences between the sexes in the ant Formica aquilonia. Gyllenstrand, N. & Seppä, P. 2003: Genetic structure in an — Heredity 10: 472–480. isolated and fragmented population of the ant Formica Pamilo, P. & Crozier, R. H. 1997: Population biology of lugubris. — Mol. Ecol. 12: 2931–2940. social insect conservation. — Memoirs of the Museum of Gyllenstrand, N., Seppä, P. & Pamilo, P. 2004: Genetic dif- Victoria 56: 411–419. ferentiation in sympatric wood ants, Formica rufa and F. Pamilo, P. & Rosengren, R. 1984: Evolution of nesting polyctena. — Ins. Soc. 51: 139–145. strategies in ants: genetic evidence from different popu- Gyllenstrand, N., Seppä, P. & Pamilo, P. 2005: Restricted lation types of Formica ants. — Biol. J. Linn. Soc. 21: gene flow between two social forms in the ant Formica 331–348. truncorum. — J. Evol. Biol. 18. [In press]. Pamilo, P. & Seppä, P. 1994: Reproductive competition Hamilton, W. D. 1964: The genetical evolution of social and conflicts in colonies of the ant Formica sanguinea. behaviour. — J. Theor. Biol. 7: 1–16. — Animal Behavior 48: 1201–1206. Hannonen, M., Helanterä, H. & Sundström, L. 2004: Habitat Pamilo, P., Chautems, D. & Cherix, D. 1992: Genetic dif- age, breeding system and kinship in the ant Formica ferentiation of disjunct populations of the ants Formica fusca. — Mol. Ecol. 13: 1579–1588. aquilonia and Formica lugubris in Europe. — Ins. Soc. Hanski, I. A. & Gilpin, M. E. 1997: Metapopulation biology: 39: 15–29. ecology, genetics and evolution. — Academic Press. Pamilo, P., Gertsch, P., Thoren, P. & Seppä, P. 1997: Molecu- Hedrick, P. W. 1999: Perspective: highly variable loci and lar population genetics of social insects. — Ann. Rev. their interpretation in evolution and conservation. — Ecol. Syst. 28: 1–25. Evolution 53: 313–318. Pamilo, P., Sundström, L. Fortelius, W. & Rosengren, R. Helanterä, H. 2004: Kinship and conflicts over male pro- 1994: Diploid males and colony level selection in For- duction in Formica ants. — Ph.D. thesis, University of mica ants. — Ethol. Ecol. Evol. 6: 221–235. Helsinki, Finland. Pamilo, P., Zhu, D. Fortelius, W., Rosengren, R., Seppä, P. Hölldobler, B. & Wilson, E. O. 1977: The number of queens: & Sundström, L. 2005: Genetic patchwork of network- An important trait in ant evolution. — Naturwissen- building wood ant populations. — Ann. Zool. Fennici schaften 64: 8–15. 42: 179–187. Hölldobler, B. & Wilson E. O. 1990: The ants. — The Pirk, C. W. W., Neumann, P., Moritz, R. F. A. & Pamilo, P. Belknap Press of Harvard University Press, Cambridge, 2001: Intranest relatedness and nestmate recognition Mass. in the meadow ant Formica pratensis. — Behav. Ecol. Keller, L. 1993: The assessment of reproductive success of Sociobiol. 49: 366–374. queens in ants and other social insects. — Oikos 67: Punttila, P. 1996: Succession, forest fragmentation, and the 177–180. distribution of wood ants. — Oikos 75: 291–298. Kilpeläinen, J., Punttila, P., Sundström, L., Niemelä, P. & Punttila, P., Haila, Y. & Tukia, H. 1996: Ant communities in Finér, L. 2005: Forest stand structure, site type and taiga clearcuts: Habitat effects and species interactions. distribution of ant mounds in boreal forests in Finland in — Ecography 19: 16–28. the 1950s. — Ann. Zool. Fennici 42: 243–258. Punttila, P., Haila, Y., Niemelä, J. & Pajunen, T. 1994: Ant Liautard, C. & Keller, L. 2001: Restricted effective queen communities in fragments of old-growth taiga and man- dispersal at a microgeographic scale in polygynous aged surroundings. — Ann. Zool. Fennici 31: 131–144. populations of the ant Formica exsecta. — Evolution 55: Queller, D. C. & Goodnight, K. F. 1989: Estimating related- 2484–2492. ness using genetic markers. — Evolution 43: 258–275. Lindström, K., Berglind, S.-Å. & Pamilo, P. 1996: Variation Ranta, E. 2002: Elinympäristön rakenteen vaikutukset puna- in colony types in the ant Formica cinerea. — Ins. Soc. kekomuurajaisten (Formica rufa) sisäsiittoisuuteen sekä 43: 329–332. kolonia- ja populaatiorakenteeseen. — M.Sc. thesis, ANN. ZOOL. FENNICI Vol. 42 • Genetic population structure and dispersal patterns in Formica ants 175

University of Helsinki, Finland. 354. Rosengren, R. & Pamilo, P. 1983: The evolution of polygyny Sundström, L. 1995a: Dispersal polymorphism and physi- and polydomy in mound-building Formica ants. — Acta ological condition of males and females in the ant, For- Ent. Fennica 42: 65–77. mica truncorum. — Behav. Ecol. 6: 132–139. Rosengren, R., Sundström, L. & Fortelius, W. 1993: Mono- Sundström, L. 1995b: Sex allocation and colony maintenance gyny and polygyny in Formica ants: the result of alter- in monogyne and polygyne colonies of Formica trunco- native dispersal tactics. — In: Keller, L. (ed.), Queen rum (Hymenoptera: Formicidae): The impact of kinship number and sociality in insects: 308–333. Oxford Uni- and mating structure. — Am. Nat. 146: 182–201. versity Press. Sundström, L., Keller, L. & Chapuisat, M. 2003: Inbreeding Ross, K. G. 1993: The breeding system of the fire ant and sex biased gene flow in the ant Formica exsecta. Solenopsis invicta: Effects on colony genetic structure. — Evolution 57: 1552–1561. — Am. Nat. 141: 554–576. Tilman, D., May, R. M., Lehman, C. L. & Nowak, M. Ross, K. G. & Shoemaker, D. D. 1997: Nuclear and mito- A. 1994: Habitat destruction and the extinction debt. chondrial genetic structure in two social forms of the fire — Nature 371: 65–66. ant Solenopsis invicta: insights into transition to an alter- Vepsäläinen, K., Savolainen, R., Tiainen, J. & Vilén, J. 2000: nate social organization. — Heredity 78: 590–602. Successional changes of ant assemblages: from virgin Rüppell, O., Heinze, J. & Hölldobler, B. 1999: Genetic and and ditched bogs to forests. — Ann. Zool. Fennici 37: social structure of the queen size dimorphic ant Lepto- 135–149. thorax cf. andrei. — Ecological Entomology 26: 76–82. Wahlund, S. 1928: Zusammensetzung von Populationen und Savolainen, R. & Vepsäläinen, K. 1988: A competition hier- Korrelationserscheinungen vom Standpunkt der Verer- archy among boreal ants: impact on resource partitioning bungslehre ausbetrachtet. — Hereditas 11: 65–106. and community structure. — Oikos 51: 135–155. Weir, B. & Cockerham, C. C. 1984: Estimating F-statistics Seifert, B. 1996: Ameisen. — Naturbuch Verlag, Augsburg. for the analysis of population structure. — Evolution 38: Seppä, P., Sundström, L. & Punttila, P. 1995: Facultative 1358–1370. polygyny and habitat succession in boreal ants. — Biol. Wright, S. 1931: Evolution in Mendelian populations. — J. Linn Soc. 56: 533–551. Genetics 16: 97–159. Seppä, P., Gyllenstrand, N., Corander, J. & Pamilo, P. 2004: Wright, S. 1943: Isolation by distance. — Genetics 28: Coexistence of the social types: Genetic population 114–138. structure in the ant, Formica exsecta. — Evolution 58: Wright, S. 1951: The genetical structure of populations. 2462–2471. — Ann. Eugenics 15: 323–354. Sundström, L. 1989: Genetic relatedness and population Wright, S. 1969: Evolution and the genetics of populations structure in Formica truncorum Fabr. (Hymenoptera, vol. 2: The theory of gene frequencies. — Univ. of Chi- Formicidae). — Actes Coll. Insectes Sociaux. 5: 93–100. cago Press, Chicago. Sundström, L. 1993: Genetic population structure and socio- Zhu, D. Q., Chapuisat, M. & Pamilo, P. 2003: Highly vari- genetic organisation in Formica truncorum (Hymenop- able social organisation of colonies in the ant Formica tera, Formicidae). — Behav. Ecol. Sociobiol. 33: 345– cinerea. — Hereditas 139: 7–12.

This article is also available in pdf format at http://www.sekj.org/AnnZool.html 176 Sundström et al. • ANN. ZOOL. FENNICI Vol. 42 . 2005 et al . 2001 . 2004 . 2004 . 2005 . 2005. . 2005 . 2005 . 2003 et al . 1997, . 2004 . 1996 et al et al et al et al et al et al . 2005 . 1992, . 1994 . 1992 . 1994 . 1994 et al . 2004 . 2004 . 2004 . 1995, . 1995 . 1995, et al et al et al . 1998, . pers. comm. . 2005 . 2003 . 2001 et al et al et al et al et al et al et al et al et al et al et al et al et al et al et al et al et al r Reference Chapuisat & Keller 1999 Pamilo & Seppä 1994 Seppä Pamilo Pirk P. Seppä unpubl. data L. Sundström unpubl. data and ha, respectively); 2 IS F

ST F

2 1 0.09* –0.001 n.a. 0.07 0.04 0.23 Mäki-Petäys Pamilo 1982 3 0.07* 0.09* 8 5 0.56 0.034* 0.03 0.16* 0.08*/0.04 Helanterä 2004 0.49/0.12 0.05 Bernasconi 2 Gyllenstrand & Seppä 2003 0.11* 0.09* 0.23 Elias 3 n.a. 2 0.02 1 n.a. 0.12* 0.04 0.0 0.03 Elias 0.01 0.15 Gyllenstrand Seppä 75 0.005 0.03/0.05 0.73/0.18 Chapuisat 50 0.004 0.08* 0.64 Seppä 50 0.15* -0.02 0.09 Seppä 15 0.05 0.14* 12 0.71 0.06* 0.005 Sundström 0.14 Mäki-Petäys 15 0.03 10 0.04 0.19* 0.02 0.47 15 0.03 Seppä 0.25 –0.09* Sundström 1993 0.67 Sundström 1993 and relatedness. The relatedness estimates are uncorrected for inbreeding. Asterisks

IS 2 4 5 4 4 4 4 50 50 15 15 1200 0.05* 100 –0.06 0.04* < –0.01 0.59 0.57 Ranta 2002 W.T. Tay, pers. comm. 36 15 400 200 n.a. 10 10 n.a. 0.17* n.a. 0.24* 0.05 ca. 100 0.04 0.07* 2–20 0.13 0.08* n.a. 0.14 n.a. n.a. Pamilo 1982, 0.01 0.22 130 Pamilo n.a. 1.5 Zhu 0.33 n.a. 15 Lindström 15 n.a. 0.03 n.a. n.a. n.a. 0.02 10–20 0.08 n.a. 0.03 160 000 n.a. 0.41 n.a. Pamilo & Rosengren 1984 ca. 10 0.04 400 0.59 n.a. n.a. 0.16* Helanterä 2004 Olsson 1999 0.47 n.a. 0.10* 200 0.24 200 n.a. 0.02* n.a. Pamilo 0.17 n.a. 0.17* Pamilo 200 ca. 1000 0.5 n.a. n.a. 400 Chapuisat n.a. 0.37 0.09 –0.04 0.16 –0.01/–0.05 0.14/0.51 0.07/0.29 Gyllenstrand 200 Beye Pamilo & Rosengren 1984 0.66 0.02* 0.63 0.12* Pamilo Pamilo 0.52 n.a. Gyllenstrand 200–1200 n.a. 0.04 n.a. 0.44 n.a. 10–15 100 Pamilo 1981, n.a. n.a. 0.07 15 0.40 0.03 –0.12* Sundström 1989, 1993, 0.60 Sundström 1993 100 50 0.09* 0.03 0.50 Seppä F Global Local Mean Mean Mean , sampling sampling 100–200* 2 0.06* 0.01 0.36 Goropashnaya ST F calculated for subsets of the entire sample. ST F = number of nests; global and local sampling are the approximate areas (in km n n Marker Area

1 112 9m Valais, Switzerland 4 17 6m Uppsala, Sweden 6 63 6m Uppsala, Sweden 7 > 100 3 9 6a 43 1 76 Uusimaa, Finland 4m 55 6 12m Uusimaa, Finland 6 Moscow, Russia ca. 60 77 1a Vihti, Finland 2a 3m Finland Europe Finland 3 2 70 4 2m+3a 71 1 Hanko, Finland 68 1-2a 1 55 3 Uusimaa, Finland 5m 34 5 5m 37 Hanko, Finland 2 74 4a Inkoo, Finland 4 12m 59 Hanko, Finland Moscow, Russia 7m 48 1 8m Peak District National Park, U.K. 2-7a 1 46 300 Swiss National Park, Switzerland Europe 21 50 4 6a 4 4m 91 Jura B, Switzerland 2 59 Jura, Switzerland > 10 5m 1 8m Siuntio, Finland 4m 2 35 Uppsala, Sweden 2 Öland, Sweden 38 4 49 2a 1-2a 8 26 Hanko, Finland 5-6a 5 Uusimaa, Finland 65 S-Finland 4m ?? 4m Siuntio, Finland 8m Inkoo, Finland Uppsala, Sweden 4 114 4 1-4a 79 S-Finland 3 4a 6 89 Hyytiälä, Finland 85 1 3a 5 15 3a Inkoo, Finland 5 30 S-Finland 1 5m 91 3-4a 2 19 Inkoo, Finland Hanko, Finland 13 3a 5m Mols, Denmark 4a Hanko, Finland Forssa N 23 249 5m Fennoscandia 11 103 3m Åland, Finland

M P

= number of populations; N the global scale as well average the population comprised monogyne and highly polygyne colonies, with very few intermediates.

Appendix. Data sources, sampling areas, sampling scales and average case; F. aquilonia F. aquilonia F. aquilonia F. candida F. cinerea F. cinerea F. cinerea F. exsecta F. exsecta F. exsecta F. exsecta F. exsecta F. fusca F. fusca F. fusca F. lugubris F. lugubris F. lugubris F. lugubris F. paralugubris F. paralugubris F. polyctena F. polyctena F. pratensis F. pratensis F. pressilabris F. rufa F. rufa F. rufa F. rufa F. sanguinea F. sanguinea F. selysi F. truncorum F. truncorum F. truncorum F. truncorum F. truncorum F. truncorum F. truncorum 1 2 indicate signiﬁcant deviations from zero at least in some populations; marker: a = allozymes, m = DNA microsatellites, the number indicates the number of loci in each

ANN. ZOOL. FENNICI Vol. 42 • Genetic population structure and dispersal patterns in Formica ants 177 . 2005 et al . 2001 . 2004 . 2004 . 2005 . 2005. . 2005 . 2005 . 2003 et al . 1997, . 2004 . 1996 et al et al et al et al et al et al . 2005 . 1992, . 1994 . 1992 . 1994 . 1994 et al . 2004 . 2004 . 2004 . 1995, . 1995 . 1995, et al et al et al . 1998, . pers. comm. . 2005 . 2003 . 2001 et al et al et al et al et al et al et al et al et al et al et al et al et al et al et al et al et al r Reference Chapuisat & Keller 1999 Pamilo & Seppä 1994 Seppä Pamilo Pirk P. Seppä unpubl. data L. Sundström unpubl. data and ha, respectively); 2 IS F

ST F

M P

= number of populations; N the global scale as well average the population comprised monogyne and highly polygyne colonies, with very few intermediates.