The evolution of agriculture in ants

Science

Volumn 281, Issue 5385, 1998, Pages 2034-2038

  Mueller, Ulrich G ^a   Rehner, Stephen A ^a   Schultz, Ted R ^a  

^a NONE

Author keywords

[No Author keywords available]

Indexed keywords

AGRICULTURE; ANT; EVOLUTION; FUNGUS; SYMBIOSIS;

AGRICULTURE; ANT; ARTICLE; CULTIVAR; EVOLUTION; FUNGUS; NONHUMAN; PHYLOGENY; POPULATION GENETIC PARAMETERS; PRIORITY JOURNAL; SYMBIOSIS;

EID: 0032566731     PISSN: 00368075     EISSN: None     Source Type: Journal    
DOI: 10.1126/science.281.5385.2034     Document Type: Article

References (25)

1. E. O. Wilson, The Insect Societies (Belknap, Cambridge, MA, 1971).
2. N. Weber, Gardening Ants (American Philosophical Society, Philadelphia, PA, 1972).
3. D. Rindos, The Origins of Agriculture (Academic Press, Orlando, FL, 1984).
4. M. Martin, Invertebrate-Microbial Interactions (Cornell Univ. Press, Ithaca, NY, 1987).
5. B. Hölldobler and E. O. Wilson, The Ants (Belknap, Cambridge, MA, 1990).
6. A. Hervey, C. T. Rogerson, I. Long, Brittonia 29, 226 (1977).
7. I. H. Chapela, S. A. Rehner, T. R. Schultz, U. G. Mueller, Science 266, 1691 (1994).
8. Although the systematics of the family Lepiotaceae is poorly understood, the lepiotaceous tribe Leucocoprini (consisting of two genera, Leucocoprinus and Leucoagaricus, and some species still assigned to Lepiota) is monophyletic [J. Johnson, thesis, Duke University, Durham, NC (1997)].
9. The tribe Attini is divided into the derived "higher attine" genera (including leafcutter ants) and the phylogenetically basal "lower attine" genera [T. R. Schultz and R. Meier, Syst. Entomol. 20, 337 (1995)]. Lower attine cultivars are largely unmodified and resemble free-living Leucocoprini, whereas higher attine cultivars are highly derived (6, 7).
10. We excluded cultivars of two derived ant clades: the subgroup of Apterostigma that cultivates non-lepiotaceous fungi, and the higher attine clade with their highly derived lepiotaceous fungi (9). The single exception is that of Trachymyrmex papulatus. Of two nests examined, one cultivated a lower attine fungus (included in Fig. 1), the other a typical higher attine fungus. This is the only known case of a higher attine ant cultivating a lower attine fungus.
11. Cultivated and free-living Lepiotaceae were surveyed in central Panama in November 1995 through December 1996. Most of the free-living fungi appear to be undescribed species ("PA" collection IDs in Fig. 1); a few collections could be associated with published descriptions ("cf." notation).
12. Our Web site (www2.sel.barc.usda.gov/Schultz/ agants.html) lists sample sizes by ant species of all 553 cultivar isolates, and additional information on phylogenetic analyses.
13. RFLPs were generated by restricting ITS polymerase chain reaction products [T. White, T. Bruns, S. Lee, J. Taylor, in PCR Protocols, M. Innis, D. Gelfand, J. Sninsky, T. White, Eds. (Academic Press, New York, 1990)] with Hae III or Taq I. Fungi were called the same type if RFLPs matched. For each ant species per collection locality, we sequenced at least one representative of each cultivated RFLP type.
14. Forward and reverse sequences were generated on an ABI 377 sequencer for the entire ITS region (680 to 740 bp) (13) and the first 610 bp of the 25S gene (7) (GenBank accession numbers AF079591-AF079754). Sequences of 11 Lepiotaceae were obtained from GenBank (U11921, U85281-U85283, U85287, U85288, U85291, U85292, U85295, U85296, U85306, U85315-U85318, U85321-U85323, U85326, U85327, U85330, U85331). We thank J. Johnson for posting these unpublished sequences in GenBank.
15. Alignments were generated in Megalign 1.1 DNA-STAR). Regions of ambiguous alignment (306 of 811 sequence positions in ITS; 5 of 611 positions in 25S) were excluded. Phylogenetic analyses were carried out in PAUP* 4.0d61 [D. Swofford, unpublished test version]. Sequence data consisted of 1422 nucleotide sites (minus the 311 unalignable characters), 229 of which were parsimony-informative. Maximum parsimony (MP) analysis identified 180 trees (length = 915, CI = 0.352, RI = 0.663). Successive approximations weighting (SAW) identified five equally parsimonious trees (all members of the original set of 180), the strict consensus of which is presented in Fig. 1. Transition:transversion weighting schemes of 1:2 and 1:5 produced parsimony trees identical in all key features to the tree in Fig. 1. To render maximum-likelihood (ML) analysis computationally tractable, we reduced the data set to 51 taxa by retaining one representative of each set of taxa differing by fewer than five bases. Each of the five SAW trees was "pruned" to retain these 51 taxa, then evaluated under the ML criterion (estimating parameters from the data). The most parameter-rich model (general time-reversible + proportion of sites invariant + rate heterogeneity modeled as a gamma distribution with six rate categories) was significantly better fitting than the next best model. Starting with this model and the most likely of these trees, and using increasingly more efficient branch-swapping algorithms and successively readjusted parameter values, five iterative ML searches identified a single most likely tree, which resembles in all key features the parsimony tree (Fig. 1). See our Web site (12) for details of the analyses.
16. A scenario of single domestication 50 million years ago, followed by escapes, is implausible because it stipulates that all free-living fungi (including major Nearctic clades) that arose after the divergence of the most recent common ancestor of all cultivars (Fig. 1) must have descended from "escaped" cultivar ancestors; it is inconsistent with observed levels of allele sequence divergence (ASD); and it contradicts the theoretical prediction that ancient clones will not retain intact the multigene architecture for fruiting [M. Lynch, R. Burger, D. Butcher, J. Hered. 84, 339 (1993)].
17. Tests contrasted unconstrained MP and ML trees with constraint trees, which were inferred using the same methods described above (15) for MP and ML, respectively. For MP, tests consisted of all possible pairwise comparisons of multiple equally parsimonious trees. Forcing cultivar monophyly (single domestication) significantly reduced goodness-of-fit [MP: Kishino-Hasegawa (KH), P < 0.0006; Templeton's Wilcoxon ranked sums (TWRS), P < 0.0007; winning sites (WS), P < 0.0003; ML: KH, P = 0.0005]. Ad hoc assumptions of two domestications involving (Clade 1 + Myrm. infuscata G11) and (Clade 2 + Myco. smithi S60) also failed (MP: KH, P < 0.0122; TWRS, P < 0.0122; WS, P < 0.0080; ML: KH, P = 0.0355). This failure was due to strong support for the monophyly of Clade 2 excluding Myco. smithi S60 (MP: KH, P < 0.0160; TWRS, P < 0.0285; WS, P < 0.0428; ML: KH, P = 0.0363). Monophyly of Clade 1 excluding Myrm. infuscata G11 was not supported (MP: KH, P > 0.4398; TWRS, P > 0.4534; WS, P > 0.3877; ML: KH, P = 0.7095). See our Web site (12) for details of the tests.
18. This firm refutation of a single domestication event suggests that further sampling may reveal additional domestications. Promising locations include Amazonian Brazil (putative center of attine origin) and peripheral populations existing under extreme ecological conditions that promote cultivar loss and prompt novel domestications.
19. It is possible that some cases of "intraspecific" cultivar diversity involve unrecognized cryptic species, each specialized on a distinct cultivar.
20. For each of six cases where different ant species farmed cultivars with identical rDNA sequences (A. auriculatum-C. longiscapus; C. minutus-C. rimosus; Myco. smithi-Myco. tardus-Myrm. ednaella; C. costatus-C. longiscapus; Myco. smithi-Myco. sp. nov. 1; Myrm. cf. buenzlii-C. faunulus) (Fig. 1), we finger-printed the sequenced isolates, and additional isolates of the same RFLP types, with AFLP techniques [U. G. Mueller, S. E. Lipari, M. G. Milgroom, Mol. Ecol. 5, 119 (1996)], permitting highly resolved subdifferentiation into AFLP fingerprint types.
21. Leucocoprinoid fungi use an ephemeral substrate (litter), are relatively short-lived, and are not known to form geographically widespread clones.
22. R. R. Snelling and J. T. Longino, in Insects of Panama and Mesoamerica, D. Quintero and A. Aniello, Eds. (Oxford Univ. Press, New York, 1992), pp. 481-494.
23. Asexual cultivar propagation was shown by AFLP fingerprint identities of cultivars from different nests of Floridan C minutus (20) and was corroborated by AFLP surveys of lower-attine cultivars from Trinidad (S. A. Rehner and U. G. Mueller, unpublished data).
24. Theory predicts elevated levels of heterozygosity (ASD) for ancient asexual diploid organisms [O. Judson and B. Normark, Trends Ecol. Evol. 11, 41 (1996)]. Without recombination, homologous alleles accumulate mutations independently and diverge with time. Heterozygosity therefore is a relative measure of the time since the origin of asexuality. Recombination (sexuality) purges ASD. Heterozygosities were scored from sequencing contigs as superimposed peaks each of about half intensity (present in forward and reverse sequences), and insertions or deletions (typically involving one base, causing a partial frame-shift at the same position, but in opposite directions, in forward and reverse sequence).
25. We thank the Smithsonian Tropical Research Institute, the National Geographic Society, the Biodiversity of the Guyanas Project, the Smithsonian Scholarly Studies Program, the Laboratory of Molecular Systematics (National Museum of Natural History), and the National Science Foundation (DEB-9707209) for funding; the Bermingham lab for sequencing support; the Instituto Nacional de Recursos Naturales Renovables (INRENARE) of Panama, the Kuna Comarca, the Office of the President of Guyana, the Ministerio de Recursos Naturales Energía y Minas (MIRENEM) and the Instituto Nacional de Biodiversidad (INBIO) of Costa Rica, the Forestry Division and Wildlife Section (FDWS) of Trinidad, and the Conselho Nacional de Pesquisas (CNPq) and the Instituto Nacional de Pesquisas da Amazónia (INPA) of Brazil for collecting and export permits; R. Adams, D. Agosti, V. Aswani, J. Boomsma, M. Braun, M. Chen, C. Currie, G. deAlba, Z. Falin, V. Funk, N. Gomez, J. Heacock, J. Huelsenbeck, J. Hunt, J. Narozniak, N. Knowlton, M. Leone, J. Longino, S. McCafferty, G. Maggiori, B. Norden, D. Piperno, I. Rubinoff, J. Sullivan, D. Swofford, J. Wilgenbusch, R. Wilson, T. Wright, and especially E. Bermingham, C. Delwiche, A. Herre, J. Kays, and B. Wcislo for logistical support, advice, and various other kinds of help. We dedicate this paper to the memory of William L. Brown Jr. (1922-1997).