Magazine R671 associations. The technology is plasticity for adopting specific suitable for long-term deployment on Wingless virgin behaviours depending on their age relatively small , and overcomes queens assume and the needs of the colony [6], key constraints of other data-collection but their reproductive potential is methods. Importantly, with its high helper roles in always permanently reduced. The sampling rates and excellent tag-to- behavioural repertoires of queens tag detection range, Encounternet can tend to be limited to mating, colony generate time-resolved association data leaf-cutting founding and egg-laying, and needed to link network topologies to this never changes in response biological processes — a key challenge to queen condition. This should in social network analyses [2,3]. Volker Nehring1,2,*, imply that gynes — unmated virgin Jacobus J. Boomsma1, queens —should be cannibalized Supplemental Information and Patrizia d’Ettorre1,3 or otherwise disposed of when Supplemental information including experi- they somehow fail to disperse on mental procedures and two figures can be Division of labour is the hallmark a mating flight, similar to infertile found with this article online at of advanced societies, because diploid males [7,8]. Uselessness for http://dx.doi.org/10.1016/j.cub.2012.06.037. specialization carries major both colony-founding and altruistic efficiency benefits in spite of behaviour means that failed queens Competing Interests Encounternet was conceived and developed costs owing to reduced individual have zero fitness and might thus at the University of Washington, Seattle, USA, flexibility [1]. The trade-off between be even expected to die voluntarily, and BO, JBu and JBo are currently working efficiency and flexibility is expressed as recycling of their tissues and on the technology’s commercial exploitation. throughout the social , where prevention of further consumption facultative social have small of scarce resources will offer them Acknowledgments colonies and reversible caste roles at least some indirect fitness when We thank C. Lambert, T. Mennesson and and advanced eusocial species their resources will benefit properly J.-E. Lombardet for facilitating field logistics, have permanently fixed queen and endowed siblings. C. Anagnostou for field assistance, W. Loo worker castes. This usually implies We observed field colonies of for help with R-code, R.C. Fleischer for that queens irreversibly specialize the leaf-cutting ants Acromyrmex molecular sexing, the BBSRC for funding on reproductive tasks [2]. Here, we echinatior and A. octospinosus that (grant BB/G023913/1 to CR), and the report an exception to this rule by contained several non-inseminated Province Sud and SEM Mwe Ara for generous showing that virgin queens (gynes) wingless gynes besides the mother support in New Caledonia. of the advanced eusocial leaf-cutting queen (Supplemental Information switch published with this article online). References 1. Krause, J., and Ruxton, G.D. (2002). Living in to carrying out worker tasks such as As these queens had been allowed Groups (Oxford: Oxford Univ. Press). brood care and colony defence when to live, we hypothesized that they 2. Whitehead, H. (2008). Analyzing Animal they fail to mate and disperse. These might still be worth their keep if Societies (Chicago: Chicago Univ. Press). 3. Croft, D.P., James, R., and Krause, J. (2008). behaviours allow them to obtain they would assume helper roles that Exploring Animal Social Networks (Princeton: indirect fitness benefits (through were of value for their colony. To test Princeton Univ. Press). 4. Rutz, C., and Hays, G.C. (2009). New frontiers assisting the reproduction of their this hypothesis, we experimentally in biologging science. Biol. Lett. 5, 289–292. mother) after their direct fitness removed the wings of normal 5. Hunt, G.R. (1996). Manufacture and use of options (their own reproduction) gynes and quantified worker-like hook-tools by New Caledonian crows. Nature 379, 249–251. have become moot. We hypothesize behaviours. As controls, we used 6. Hunt, G.R., and Gray, R.D. (2003). that this flexibility could (re-)evolve both completely un-manipulated Diversification and cumulative evolution in New Caledonian crow tool manufacture. Proc. secondarily because these ants only gynes, and gynes with one middle R. Soc. B 270, 867–874. feed on fungal mycelium and thus leg removed. This mutilation was not 7. Rutz, C., Bluff, L.A., Weir, A.A.S., and Kacelnik, could not benefit from cannibalising expected to affect normal behaviour A. (2007). Video cameras on wild birds. Science 318, 765. redundant gynes, and because as queens missing a leg are 8. Laland, K.N. (2008). Animal cultures. Curr. Biol. queens have retained behavioural occasionally observed as functional 18, R366–R370. 9. Holzhaider, J.C., Sibley, M.D., Taylor, A.H., repertoires for foraging, nursing, colony mothers. Singh, P.J., Gray, R.D., and Hunt, G.R. (2011). and defense, which they naturally Acromyrmex gynes with their The social structure of New Caledonian crows. express during colony founding. wings experimentally removed were Anim. Behav. 81, 83–92. 10. Whitehead, H., and Lusseau, D. (2012). Animal Permanence of caste is the most more likely to conduct worker tasks: social networks as substrate for cultural fundamental difference between unlike the control gynes, they took behavioural diversity. J. Theor. Biol. 294, 19–28. eusocial and cooperative breeding care of brood that we introduced 1Department of Zoology, University of Oxford, [2]. Helpers of the latter category into experimental nests (Figure 1A,B) South Parks Road, Oxford OX1 3PS, UK. may become breeders later in and they attacked non-nestmates 2Present address: School of Biology, University life and thus retain full flexibility (Figure 1C). A second nestmate of St Andrews, St Andrews KY16 9TH, UK. towards the eventual pursuit of recognition experiment, in which 3Department of Physics, University of Bath, direct fitness. Illustrative examples tethered gynes were presented with 4 Bath BA2 7AY, UK. School of Biological are cooperatively breeding birds, nestmate or non-nestmate odour, Sciences, University of Auckland, PB 92019, Auckland 1142, NZ. 5Department of Electrical paper wasps and some Ponerinae showed that both the control and Engineering, University of Washington, Seattle, ants [3–5]. Workers of eusocial wingless gynes were equally able to WA 98195-3770, USA. ants and honeybees, by contrast, discriminate between nestmate and *E-mail: [email protected] at best retain some phenotypic non-nestmate cues (Figure 1D). Current Biology Vol 22 No 17 R672

is not an oversight by generations of natural historians, but that there are likely to be two selection forces that make working/defending gynes particularly probable in fungus- growing ants such as Acromyrmex. First, Acromyrmex queens are semiclaustral, which means that they forage during colony founding, unlike the claustral queens of most ants that seal their founding burrow to live on body reserves until their first workers hatch [9]. Acromyrmex gynes have thus retained a working and defending repertoire that becomes temporarily expressed in response to dispersal and insemination, so could easily be co-opted. Second, attine ants consume almost exclusively fungal , and are thus likely to have lost the ability to digest animal tissue. Acromyrmex fungus-gardens have also never been observed to contain parts of dead insects as substrate in any appreciable amount [10]. This means that workers cannot recoup investments that went into raising reproductives, so they may as well let wingless gynes serve the colony while living on the fat and carbohydrate reserves that Figure 1. Virgin queens (gynes) as helpers in Acromyrmex leaf-cutting ants. were deposited to prepare them Wingless Acromyrmex gynes carry out worker behaviour in contrast to winged control gynes for founding a new colony. This (leg-mutilated and un-manipulated). (A and B) They transported brood items back to their would imply that wingless gynes fungus garden (three colonies tested) and (C) attacked non-nestmate intruders, while control expressing worker behaviour are gynes rarely engaged in these behaviours (boxplots show interquartile ranges and medians; not expected to be fed, because whiskers cover all data points; four colonies tested in the field and laboratory, see also Figure their existence only makes sense S1). Gynes hardly ever attacked nestmates (p < 0.001, interaction gyne type x intruder p < 0.001). (D) Tethered wingless gynes and control gynes were all able to discriminate non-nest- when their siblings profit at no mates from nestmates: all three types of gynes were more prone to opening their mandibles cost. However, our current set-up (showing threat) when confronted with non-nestmate than with nestmate odour or solvent (p < did not allow us to investigate this 0.05, three colonies). Wingless gynes were generally more likely to open their mandibles than possibility. other gynes (p < 0.05), consistent with expression of more aggressive behaviour; the ability to We hypothesize that the same discriminate between nestmate and non-nestmate odours was, however, similar for all gynes logic may apply in other attine ants (interaction odour x gyne type p > 0.05). Sample sizes are indicated above the bars and boxes (*** p < 0.001, * p < 0.05, n.s. p > 0.05). and that soldier/nurse gynes might occur also in related higher attine genera such as Trachymyrmex This indicates that the lack of As comparative context for our and Sericomyrmex that rear very aggression in control gynes is not experimental results we provide a similar fungal symbionts. However, due to their inability to recognise summary of cases where ant queens we expect that such gynes will intruders (differential perception have been reported to perform never be found in , as this of recognition cues), but to gynes worker behaviour (Supplemental secondarily evolved claustral avoiding intruders (differential Information). They all concerned ants colony founding, so that gynes in decision rule) when dispersal and with either many queens per colony any condition may no longer have reproductive potential is intact. or special gyne-morphs that can a worker-like behavioural repertoire Only after the option to disperse later become replacement queens to activate either after or before and mate was lost did gynes when they mate in the nest, but no mating. Consistent with this idea, switch to the more risky behaviour case reminiscent of Acromyrmex we have never observed wingless of attacking intruders, a task that where colonies are predominantly queens when digging out mature their larger body size allows them (A. echinatior) or almost exclusively Atta colonies but we have regularly to do more effectively than large (A. octospinosus) headed by a single observed Atta colombica workers workers (Acromyrmex never evolved queen and where queens always dumping dead reproductives a soldier caste as their sister genus have a dispersal flight and never on their above ground compost Atta did). mate in the nest. We believe that this heaps shortly after a mating flight, Magazine R673 confirming that their resources printed in multi-well glass-bottomed cannot be recycled. The first World Cell cell-culture wells (see Supplemental Race Experimental Procedures and Supplemental Information Figure S1A in the Supplemental Supplemental Information including experi- Information available online); 54 mental procedures, a figure and a table can Paolo Maiuri1, Emmanuel Terriac1, different cell types from various be found with this article online at Perrine Paul-Gilloteaux1, animals and tissues were provided http://dx.doi.org/10.1016/j.cub.2012.064.038. Timothée Vignaud2, Krista McNally3, by 47 laboratories. Genotypically, James Onuffer3, Kurt Thorn3, cells were wild type, transformed or Acknowledgments Phuong A. Nguyen4, genetically engineered (Table S1). We thank The Smithsonian Tropical Nefeli Georgoulia4, Daniel Soong5, The cells were distributed to six Research Institute for providing facilities Asier Jayo5, Nina Beil6, organizing laboratories (two in the in Gamboa that allowed us to collect ant Jürgen Beneke6, USA, and one each in the UK, France, colonies and conduct the behavioural Joleen Chooi Hong Lim7, Germany and Singapore), who assays, Luke Holman and two anonymous Chloe Pei-Ying Sim7, prepared cell-culture stocks using reviewers for helpful comments on an Yeh-Shiu Chu7, WCR participants8, the frozen samples received from earlier version of the manuscript, and Andrea Jiménez-Dalmaroni9, participating laboratories and plated the members of the Centre for Social Jean-François Joanny9, these onto the race tracks under Evolution for creating a pleasant working Jean-Paul Thiery7, Holger Erfle6, identical culture conditions. Cells environment. This study was supported by Maddy Parsons5, were allowed to adhere overnight The Danish National Research Foundation Timothy J. Mitchison4, and cell was recorded for (J.J.B.), the German Academic Exchange 3 Service DAAD (V.N.) and a Freia grant Wendell A. Lim , 24 hours using an inverted video 10 from the Faculty of Science, University of Ana-Maria Lennon-Duménil , microscope (Figure 1B, Movie S1). 1, Copenhagen (P.d.E.). Matthieu Piel *, Cell morphology (length, shape, and Manuel Théry2,* symmetry, and nucleus position) varied greatly from one cell type References Motility is a common property to another (Figure 1A). Cell nuclei 1. Strassmann, J.E., and Queller, D.C. (2007). societies as divided organisms: The of animal cells. Cell motility is were stained by incubating live complexities of purpose and cross-purpose. required for embryogenesis [1], cells with 5 ng/ml Hoechst dye Proc. Natl. Acad. Sci. USA 104, 8619–8626. 2. Boomsma, J.J. (2009). Lifetime monogamy tissue [2] and the diluted in normal growth medium. and the evolution of . Phil. Trans. immune response [3] but is also Cell displacements were monitored Biol. Sci. 364, 3191–3207. involved in disease processes, such every 10 minutes. Nuclei images 3. Cornwallis, C.K., West, S.A., Davis, K.E., and Griffin, A.S. (2010). Promiscuity and the as of cancer cells [4]. were segmented and geometric evolutionary transition to complex societies. Analysis of cell migration in native centers were tracked with a global Nature 466, 969–972. 4. Leadbeater, E., Carruthers, J.M., Green, tissue in vivo has yet to be fully minimization algorithm in order to J.P., Rosser, N.S., and Field, J. (2011). Nest explored, but motility can be relatively track automatically individual cell inheritance is the missing source of direct easily studied in vitro in isolated displacements (see Supplemental fitness in a primitively eusocial insect. Science 333, 874–876. cells. Recent evidence suggests that Experimental Procedures and Figure 5. Hart, A.G., and Ratnieks, F.L.W. (2005). cells plated in vitro on thin lines of S1B). The motility of over 7,000 cells Crossing the taxonomic divide: conflict and its resolution in societies of reproductively adhesive proteins printed onto culture was compared, with an average of totipotent individuals. J. Evol. Biol. 18, dishes can recapitulate many features 130 cells analyzed per cell type. 383–395. of in vivo migration on collagen Detailed statistical analyses were 6. Gordon, D.M. (1996). The organization of work in social insect colonies. Nature 380, fibers [5,6]. However, even with used to characterize cell motility 121–124. controlled in vitro measurements, parameters for each cell type (see 7. Nielsen, M.G. (1978). Production of sexuals in nests of Lasius flavus (Forst.) (: the characteristics of motility are http://www.worldcellrace.com/ Formicidae). Nat. Jutl. 20, 251–254. diverse and are dependent on the ResultFiles). 8. Dijkstra, M.B., and Boomsma, J.J. (2007). cell type, origin and external cues. The mean instantaneous speed The economy of worker reproduction in Acromyrmex leafcutter ants. Anim. Behav. One objective of the first World Cell of individual cells is computed by 74, 519–529. Race was to perform a large-scale averaging the cell displacements 9. Keller, L. (1991). Queen number, mode of colony founding, and queen reproductive comparison of motility across many between consecutive frames over success in ants (Hymenoptera, Formicidae). different adherent cell types under time. The distribution of mean Ethol. Ecol. Evol. 3, 307–316. standardized conditions. To achieve instantaneous speeds for each cell 10. De Fine Licht, H.H., and Boomsma, J.J. (2010). Forage collection, substrate a diverse selection, we enlisted type was asymmetric (Figure 1C) preparation and diet composition in fungus- the help of many international and non-Gaussian (Figure S1C). growing ants. Ecol. Entomol. 35, 259–269. laboratories, who submitted cells for Interestingly, we observed that a analysis. The large-scale analysis, higher mean speed for a given cell 1Centre for Social Evolution, Department made feasible by this competition- type did not reflect a global shift of of Biology, University of Copenhagen, oriented collaboration, demonstrated the speed distribution, but rather the 2100 Copenhagen, Denmark. 2Biology that higher cell speed correlates with spreading of the distribution due to I, University of Freiburg, Hauptstraße 1, the persistence of movement in the the presence of faster moving cells 3 79104 Freiburg, Germany. Laboratoire same direction irrespective of cell (Figure 1C and Figure S1C). In order d’Ethologie Expérimentale et Comparée (LEEC), Université Paris 13, 93430 origin. to identify the 2011 World Cell Race Villetaneuse, France. The race track consisted of 4 mm- winner, only cells with an effective *E-mail: [email protected] and 12 mm-wide lines overall displacement of at least