4.3 Brief Comms Am 26/2/04 6:00 Pm Page 1

4.3 brief comms am 26/2/04 6:00 pm Page 1

brief communications Ant parasite queens revert to mating singly A parasitic ant has abandoned the multiple mating habit of the queens of its related host.

ultiple mating (polyandry) is would apply equally to social parasites widespread among animal and free-living species. Other likely ASH 1 Mgroups, particularly insects .But benefits (greater sperm store, conver- D. R. N the factors that maintain it and under- gence of sex-allocation interests lie its evolution are hard to verify between queens and workers, reduced because benefits and costs are not easily fitness cost of producing sterile diploid quantified and they tend to be similar males and increased genetic diversity in related species. Here we compare the among workers3) are specific to free-liv- mating strategies of the leaf-cutting ant ing social insects and so are the most Acromyrmex echinatior and its recently likely explanations for polyandry in derived social parasite Acromyrmex Acromyrmex ants. The benefits of insinuator,which is also its closest rela- polyandry in these ants are therefore tive2 (see Fig. 1). We find that although related to their social lifestyle, indicat- the host queens mate with up to a ing that the selective forces affecting the dozen different males, the social para- evolution of polyandry in free-living site mates only singly. This rapid and Figure 1 A multiply mated queen of the leaf-cutting ant Acromyrmex echinatior social insects differ from those in other surprising reversion to single mating in (right) and its singly mated socially parasitic companion queen A. insinuator (left) animals. We would expect workerless a socially parasitic ant indicates that the on their joint fungus garden. Small worker ants cluster around the parasite queen. social parasites of other polyandrous costs of polyandry are probably specific social insects to have also reverted to to a free-living lifestyle. P0.0006), and genetic relatedness among mating singly. The identification of addi- In social insects, obligate polyandry has the offspring was 0.3800.042 versus tional reversions to single mating, both in evolved independently in four groups 0.6930.053 (t-test, P0.0002) for the host social and non-social animals,will be impor- (honeybees,vespid wasps,harvester ants and and parasite,respectively.The difference can- tant in further distinguishing the factors leaf-cutting ants3), among which every not be attributed to low variability in the species was previously believed to be multi- genetic markers8,sampling error8 or inbreed- ply mated. All free-living leaf-cutting ants ing: neither species was inbred (F0.006, 4 are multiply mated , so the common ances- n10, P0.19 and F0.035, n19, 48891010 10 11 11 12 1.0 tor of both parasite and host ants must have P0.88 for A. echinatior and A. insinuator, had multiply mated queens. Because social respectively). 0.8 Host parasite queens exploit the resources and These findings provide insight into the 0.6 workers of a host colony to produce repro- selective forces affecting the evolution of 0.4 ductive offspring (sexuals) without investing polyandry. Like its host, A. insinuator is out- 0.2 in a large worker force of their own5–7,we bred and produces hundreds of winged sex- 0 investigated whether this dramatic change in uals in discrete batches that leave their natal 7 1111111111122 lifestyle could have altered the costs and colonies within the same brief time period . 1.0 benefits of polyandry, and thereby driven a This indicates that females of both species 0.8 shift in mating strategy. mate with unrelated males in swarms outside 0.6 Parasite Parasitized A. echinatior colonies were the colony. The parasite is also very abun- 0.4 collected from Gamboa, Panama, where the dant, infecting 44% of host colonies at our Frequency of patrilines within progeny 0.2 species is common7.We compared the mat- study site7.Mating behaviour,mate availabil- ing frequencies of host and parasite by ity and thus the costs of mating are therefore 0 analysing the genotypes of queens and their likely to be similar for the two species. That Queens female offspring to estimate the number of A. insinuator has reverted to single mating fathers contributing to a queen’s progeny. suggests that the benefits of polyandry are Figure 2 The observed frequency distributions of patrilines Host queens mated with 10.61.0 males reduced for the social parasite.The reversion (offspring sired by different males) of the social parasite (means.e; n10) but social parasite to single mating seems to have been rapid Acromyrmex insinuator (n13) and its host A. echinatior queens were predominantly singly mated compared with the evolution of other adap- (n10), as derived from genetic analysis of the progeny of (average number of matings per queen, tive traits5–7,supporting the widely held but single queens. Patrilines are shown by alternate shading patterns, 1.190.10; only 2 out of 13 queens doubly rarely demonstrated assumption1,2 that with their total number in each colony given above the bars. Host mated;see Fig.2). multiple mating is costly. queens and an average of 203 (range, 35–672) female offspring This difference is statistically highly sig- The association between a reversion to per queen were genotyped at four microsatellite loci (Ech1390, nificant (Mann–Whitney U-test,P0.0001) single mating and a switch to social para- Ech3385, Ech4126 and Ech4225)9.Parasite queens and an and robust to alternative measures of queen- sitism indicates that the benefits of average of 14.7 (6–20) female progeny per queen were geno- mating frequency. For example, the observed polyandry for Acromyrmex ants may be spe- typed at two loci (Trachy11_12 and Ech3385)9,10. Genotypes of number of queen matings was 9.30.7 cific to being free-living and social.Four pro- queens and their offspring were analysed using the software versus 1.150.10 (P0.0001), the effective posed benefits of multiple mating for social MATESOFT 1.0b (by A. Moilanen, L. Sundström and J. S. P.; avail- number of queen matings was 5.30.6 insect queens (nutrient transfer,sperm com- able at http://www.zi.ku.dk/popecol/personal/JSPedersen/Mate- versus 1.090.09 (P0.0001),the estimated petition, variable daughter queens and Soft.htm) to obtain mating-frequency statistics8, and the total proportion of multiply mated queens was avoidance of genetic incompatibility3) are mate number for host queens was further estimated using a 1.00 (95% confidence interval, 0.69–1.00) unlikely to be important in the maintenance modified version of the coverage statistics calculated by the versus 0.19 (0.034–0.51; binomial test, of polyandry in Acromyrmex,because they program CHAO11.

brief communications

involved in the evolution and maintenance the basis of previous correlations with 3 of multiple mating by females. spreading rate” .Owing to the high enrich- a Seirian Sumner*, William O. H. Hughes, ment of the hydrothermal fluids in mantle 25

3 4 )

Jes S. Pedersen, Jacobus J. Boomsma –1 He (typically a factor of 10 relative to ambi- 20

Biological Institute, Department of Population ent sea water), this abundant hydrothermal km Biology, University of Copenhagen, activity should logically correspond to an –1 15 3 yr Rainbow –1 2100 Copenhagen, Denmark enhanced He output from the ridge. Lucky Strike e-mail: [email protected] At a global scale, three-dimensional 10 Red Sea *Present addresses: Smithsonian Tropical Research oceanic general circulation models make a 5 He (mmol Institute, Unit 0948, APO AA 34002, USA, and good job of simulating the oceanic distribu- 3 Gakkel Ridge Institute of Zoology, Zoological Society of London, tion of hydrothermal 3He by using the simple 0 3 010203040506070 Regents Park, London N1 4RY, UK. assumption that the He source intensity is Spreading rate (mm per year) 1. Simmons, L. W. Sperm Competition and its Evolutionary linearly correlated with the spreading rate5,6. Consequences in the Insects (Princeton University Press, 3 b Princeton, 2001). The slope of the He flux–spreading rate rela- Barents Sea 3 2 2. Sumner, S., Aanen, D. K., Delabie, J. & Boomsma, J. J. Insectes tionship is close to 330 He mol km of 39 ± EBDW+EBDW Ar: 92 6% Soc. 51, 37–42 (2004). crustal spreading, translating into a world 3He: 0.6±0.3% 3. Crozier, R. H. & Fjerdingstad, E. J. Ann. Zool. Fenn. 38, 267–285 3 2 (2001). average He annual flux of 0.33 mmol km 39Ar: 60.1±6.6% 4. Villesen, P., Murakami, T., Schultz, T. R. & Boomsma, J. J. of ridge and per mm yr1 of full spreading Proc. R. Soc. Lond. B 269, 1541–1548 (2002). 3He: 4.1±0.7% NSDW 3 5. Sumner, S., Nash, D. R. & Boomsma, J. J. Proc. R. Soc. Lond. B rate.Available He fluxes determined for 39Ar: 83±5%

270, 1315–1322 (2003). active portions of ridge show results consis- 3He: 1.9±0.6% 6. Sumner, S., Hughes, W. O. H. & Boomsma, J. J. Behav. Ecol. Sociobiol. 54, 256–263 (2003). tent with this global trend (Fig.1a). 7. Bekkevold, D. & Boomsma, J. J. J. Evol. Biol. 13, 615–623 (2000). Gakkel Ridge stretches 1,800 km across the Figure 1 3He fluxes at mid-ocean ridges. a,3He annual flux per 8. Pedersen, J. S. & Boomsma, J. J. Mol. Ecol. 8, 577–587 (1999). kilometre of ridge as a function of spreading rate. The linear rela- 9. Ortius-Lechner, D., Gertsch, P. J. & Boomsma, J. J. Mol. Ecol. 9, Eurasian basin of the Arctic Ocean.Residence 114–116 (2000). time of deep and bottom waters in the Green- tionship used in GCM simulations5,6 is shown (dashed line). The 10.Villesen, P., Gertsch, P. J. & Boomsma, J. J. Mol. Ecol. Notes 2, land/Norwegian seas and the Eurasian basin Mid-Atlantic Ridge (MAR) Lucky Strike site and Red Sea values are 320–322 (2002). 11.Chao, A. & Lee, S. M. J. Am. Stat. Assoc. 87, 210–217 (1992). have been studied in detail in the recent past from refs 9,10. The Rainbow site (MAR) value is unpublished data. Competing financial interests: declared none. using multi-tracer balances7. 3He concentra- b, Eurasian basin box model : Eurasian basin deep and bottom tions are relatively low in the deep waters of waters (EBDW and EBBW respectively) are directly ventilated by the the Arctic Ocean: Eurasian basin deep waters sinking of dense shelf waters from the Barents Sea. They leave the COMMUNICATIONS ARISING (EBDW), which are at 1,500–2,600 m depth, basin through the western part of the Fram Strait and mix with the Arctic Ocean show a 3He excess relative to solubility equilib- Greenland Sea and Norwegian Sea deep waters (NSDW). In addi- rium of 4.30.8%, whereas the Eurasian tion to the Barents Sea waters, some NSDW also return to the Hydrothermal activity on basin bottom waters (EBBW) average 3He north and re-enter the deep Arctic through the eastern part of the Gakkel Ridge excess is 3.70.6% (refs 7,8). Fram strait7. 39Ar values7 are expressed as a percentage of modern The 3He input from Gakkel Ridge to the concentration. 3He data are expressed using the isotope notation. In the hydrothermal circulation at mid- EBBW and EBDW can be calculated using a ocean ridges, sea water penetrates the frac- simplified version of a box model7 (Fig.1b). fewer chemicals and less heat, so higher fre- tured crust, becomes heated by its proximity Water fluxes are derived from the 39Ar quency does not necessarily mean higher to the hot magma, and returns to the sea radioactive tracer budget (39Ar is a cosmo- hydrothermal activity. The apparent discrep- floor as hot fluids enriched in various chemi- genic isotope with a half-life of 269 years), ancy between the enhancement factor of two cal elements. In contradiction to earlier allowing the mantle 3He flux to be calculated to three in the vent site frequency deduced results1 that predict diminishing hydrother- from the 3He budget.Taking into account the from along axis tracer surveys and the low 3He mal activity with decreasing spreading rate, a uncertainties in the data, a maximum 3He flux may therefore primarily reflect the survey of the ultra-slowly spreading Gakkel input of 3.4mol yr1 is inferred.Scaled to the unusual characteristics of the hydrothermal Ridge (Arctic Ocean) by Edmonds et al.2 and length and spreading rate of Gakkel Ridge, circulation on this ultra-slowly spreading Michael et al.3 suggests that, instead of being the mantle 3He input translates into an ridge. rare, the hydrothermal activity is abundant annual flux of 0.15–0.32 mmol km1 of ridge Detailed investigation of individual vent- — exceeding by at least a factor of two to and per mm yr1 of full spreading rate ing sites in such environments is needed to three3 what would be expected by extrapola- (Fig.1a), showing no positive anomaly with resolve this issue and to reassess the geologi- tion from observation on faster spreading respect to that expected from the value of the cal parameters that control ridge processes, ridges. Here we use helium-3 (3He), a spreading rate. including hydrothermal plume and vent fre- hydrothermal tracer4,to show that this abun- One explanation mentioned by Michael quency,fluid circulation and fluxes. dance of venting sites does not translate, as et al.3 for the observed discrepancy could be Philippe Jean-Baptiste, Elise Fourré would be expected, into an anomalous that the high plume incidence observed dur- Laboratoire des Sciences du Climat et de hydrothermal 3He output from the ridge. ing the survey (80% plume incidence on the l’Environnement, IPSL, CEA-CNRS, Centre Because of the wide implications of the sub- 145 profiles) overestimated the abundance d’Etudes de Saclay, 91191 Gif-sur-Yvette, France marine hydrothermal processes for mantle of vents because of the confined nature of the e-mail : [email protected] heat and mass fluxes to the ocean, these con- deep unsegmented rift valley. It is possible 1. Baker, E.T., Chen, Y. J. & Phipps Morgan, J. Earth Planet. Sci. Lett. 142, 137–145 (1996). flicting results call for clarification of the link that plume meanders originating from a sin- 2. Edmonds, H. N. et al. Nature 421, 252–256 (2003). between hydrothermal activity and crustal gle vent site may have been interpreted as 3. Michael,P.J.et al. Nature 423, 956–961 (2003). production at mid-ocean ridges. corresponding to different sources. Alterna- 4. Craig, H.& Lupton,J.E.in The Sea Vol.7, 391–428 (Wiley, New York, 1981). On the basis of the axial distribution of tively, if the inferred number of vent sites is 5. Farley,K. A., Maier-Reimer, E., Schlosser, P. & Broecker, W. S. light scattering and manganese tracer anom- right,an explanation may rest on the fact that J. Geophys. Res. 100, 3829–3839 (1995). 6. Dutay, J.C. et al. J. Mar. Syst. (in the press). alies measured during the Arctic mid-ocean vent frequency is not a direct measurement 7. Boenisch, G. & Schlosser, P. Prog. Oceanogr. 35, 29–52 (1995). ridge experiment (AMORE), hydrothermal of hydrothermal activity. 8. Schlosser, P. et al. Prog. Oceanogr. 35, 1–28 (1995). 9. Jean-Baptiste, P. et al. Earth Planet. Sci. Lett. 157, 69–77 (1998). activity on Gakkel Ridge has been reported It is also possible that, on ultra-slowly 10.Jean-Baptiste, P., Fourré, E., Metzl, N., Ternon, J. F. & Poisson, to be “far more abundant than predicted on spreading ridges, individual plumes carry A. J. Mar. Syst. (in the press).