Ant Species Identity has a Greater Effect than Fire on the Outcome of an Ant Protection System in Brazilian Cerrado Kleber Del-Claro 1,3 and Robert J. Marquis 2 1 Instituto de Biologia, Universidade Federal de Uberl^ andia (UFU), CP 593, Uberl^ andia, MG, CEP 38400-902, Brazil 2 Department of Biology and Whitney R. Harris World Ecology Center, University of Missouri-St. Louis, One University Boulevard, St. Louis, MO, 63121, U.S.A. ABSTRACT Although re- and antplant interactions inuence the community structure and dynamics of Neotropical savannas, no previous studies have considered their simultaneous effects on target host plants. We monitored the effect of ant exclusion for 3 years on leaf area loss to leaf chewing insects, thrips abundance, and reproductive output of the extraoral nectary-bearing shrub, Peixotoa tomentosa (Malpighia- ceae). We predicted that the impact of ants on herbivores and plants would depend on the ant species, and that re would reduce the effect of ants. We deliberately chose control plants that differed in their occupant ant species. Fire occurred in the second year of the study, allowing us to determine its effect on the benet afforded by ants. Ants reduced leaf area loss and thrips abundance, and increased fruit and seed production in all 3 years. Some ant species were more effective than others, while plants with multiple ant spe- cies suffered higher leaf area loss than plants with a single ant species. In the year following the re, leaf damage was greater than in the other years, regardless of the ant species, and the proportional effect of ants in reducing damage was less. Interactions affecting thrips abundance did not change following re, nor was the benet to the plant proportionally reduced. Overall, the identity of the ant species had a greater effect than did the occurrence of re on the antherbivoreplant interaction: the identity of the ant species inu- enced leaf area loss, thrips numbers, and bud and seed production, while re only modied the impact of ants on the amount of leaf area consumed by insect herbivores. Key words : abiotic factors; antplant interactions; Brazil; Cerrado; extraoral nectaries; re; thrips. EXTRAFLORAL NECTARIES (EFNS), WHICH ATTRACT ANTS AND OTHER INVERTEBRATES THAT FEED ON THEIR SECRETIONS (Byk & Del-Claro 2011), have been present in Angiosperms since the Oligocene (Pemberton 1992). Ants visiting EFN-containing plants often protect those plants by eating, hindering, or driving away herbi- vores (Bentley 1977, Horvitz & Schemske 1984, Koptur 1984, Vilela et al. 2014). In turn, plant growth and seed production are often higher (Del-Claro et al. 1996, Oliveira 1997, De la Fuente & Marquis 1999, Nascimento & Del-Claro 2010). Not all studies, however, suggest that EFNs are benecial for plants (ODowd & Catchpole 1983, Rashbrook et al. 1992, Bronstein 1998, Mody & Linsenmair 2004, Chamberlin and Hol- land 2009). Differences in behavior among the ant species attracted to plants containing EFNs inuence the effect of the EFNs (Del-Claro & Oliveira 2000, Mody & Linsenmair 2004, Palmer et al. 2008, Nahas et al. 2012, Del-Claro et al. 2013). The abiotic environment may also contribute, although this is much less studied (Bronstein 1998, Vilela et al. 2014, Yamawo et al. 2014). Clarifying which factors inuence the effect of EFNs is an important starting point for understanding how these antherbi- voreplant interactions evolved and were maintained, and for identifying their contribution to the structure and functioning of food webs. This is particularly true for tropical systems where plant species with EFNs are most abundant (Rico-Gray & Olive- ira 2007, Bronstein 2009, Dattilo et al. 2014). Plants with EFNs are common in the savannas of South America, Africa, and Asia (Rico-Gray & Oliveira 2007). Because re is such an important environmental factor in savanna vege- tation (Alves-Silva & Del-Claro 2014), it is likely that re, in addition to ant species identity, inuences the nature of interac- tions between ants, herbivores, and plants with EFNs. In Brazil- ian savanna (the Cerrado), EFNs occur on more than 25 percent of the tree species and up to 30 percent of individual trees (Rico-Gray & Oliveira 2007). Fire is considered to be a principal determinant of Cerrado vegetation structure and plant diversity (Miranda et al. 2002). One obvious mechanism by which re might inuence the interactions between plants and ants visiting EFNs is through its impact on insect herbivores and ants. Cerrado arthropod density and ant species richness are known to decline in re-disturbed areas as a result of direct mortality, diminished resources, and less favorable microclimates (Vasconcelos et al. 2009). Although both antplant interactions (Del-Claro & Torezan-Silingardi 2009) and re (Vasconcelos et al. 2009) play an important role in the community structure and dynamics of Cerrado (Alves-Silva & Del-Claro 2014), their simultaneous effects on plants in any tropical savanna system are unknown. Received 9 June 2014; revision accepted 26 February 2015. 3 Corresponding author; e-mail: delclaro@ufu.br ª 2015 The Association for Tropical Biology and Conservation 459 BIOTROPICA 47(4): 459–467 2015 10.1111/btp.12227