344 Biological Journal of the Linnean Society, 2018, 125, 344–354. With 3 figures. © 2018 The Linnean Society of London, Biological Journal of the Linnean Society, 2018, 125, 344–354 Ant–plant relationships in the canopy of an Amazonian rainforest: the presence of an ant mosaic ALAIN DEJEAN 1,2 *, JÉRÔME ORIVEL 2 , MAURICE LEPONCE 3,4 , ARTHUR COMPIN 1 , JACQUES H. C. DELABIE 5 , FRÉDÉRIC AZÉMAR 1 and BRUNO CORBARA 6 1 Ecolab, Université de Toulouse, CNRS, Toulouse, France 2 CNRS, UMR EcoFoG, AgroParisTech, Cirad, INRA, Université des Antilles, Université de Guyane, 97310 Kourou, France 3 Biodiversity Monitoring & Assessment, Royal Belgian Institute of Natural Sciences (RBINS), Rue Vautier 29, B-1000 Brussels, Belgium 4 Behavioural & Evolutionary Ecology, Université Libre de Bruxelles (ULB), Av. F.D. Roosevelt 50, B-1050 Brussels, Belgium 5 U.P.A. Laboratório de Mirmecologia, Convênio UESC/CEPLAC, C.P. 7, 45600-000 Itabuna, Bahia, Brazil 6 Université Clermont Auvergne, CNRS, LMGE, F-63000 Clermont-Ferrand, France Received 11 July 2018; accepted for publication 2 August 2018 Using different techniques to access the canopy of an Amazonian rainforest, we inspected 157 tree crowns for arbo- real ants. Diversity statistics showed that our study sample was not representative of the tree and ant populations due to their high diversity in Amazonian rainforests, but permitted us to note that a representative part of territori- ally dominant arboreal ant species (TDAAs) was inventoried. Mapping of TDAA territories and use of a null model showed the presence of an ant mosaic in the upper canopy, but this was not the case in the sub-canopy. Among the TDAAs, carton-nesting Azteca dominated (52.98% of the trees) whereas ant-garden ants (Camponotus femoratus and Crematogaster levior), common in pioneer formations, were secondarily abundant (21.64% of the trees), and the remaining 25.37% of trees sheltered one of 11 other TDAAs. The distribution of the trees forming the upper canopy influences the structure of the ant mosaic, which is related to the attractiveness of some tree taxa for certain arboreal ant species and represents a case of diffuse coevolution. ADDITIONAL KEYWORDS: ant mosaics – host tree attractiveness – indicators of disturbance – pristine Amazonian rainforest – territoriality. INTRODUCTION Interspecific interactions can be neutral, antagonistic (i.e. competitive, predatory or parasitic) or mutualis- tic, resulting in complex webs of interacting species that influence their evolution and diversification (for mutualisms see Bascompte & Jordano, 2013; Knight et al., 2018). The Red Queen Hypothesis (Van Valen, 1973) posits that organisms interacting antagonisti- cally in a given community must evolve constantly to survive and gain a reproductive advantage and, so, are involved in an ‘evolutionary arms race’ (see Dyer, 2014). This Red Queen effect was also shown for mutu- alisms, more exactly for tight associations between ants and myrmecophytes (i.e. plants sheltering colo- nies of a limited number of ant species in hollow struc- tures) (Rubin & Moreau, 2016). Thus, in addition to their adaptations to abiotic fac- tors, populations evolved in a context of large networks of interacting species that become mutually adapted (increased fitness). This ‘diffuse coevolution’ contrasts with reciprocal evolutionary changes between two spe- cies, or ‘pairwise coevolution’ (Janzen, 1980) that, cor- responding to specialization, is the exception (Fortuna et al., 2013). Pairwise coevolution is seen in a few myr- mecophyte species but only at the regional scale (see *Correspondence: alain.dejean@wanadoo.fr Downloaded from https://academic.oup.com/biolinnean/article/125/2/344/5090152 by guest on 26 December 2023