Thresholds in the relationship between functional diversity and patch size for mammals in the Brazilian Atlantic Forest M. Magioli 1 , M. C. Ribeiro 2 , K. M. P. M. B. Ferraz 1 & M. G. Rodrigues 3 1 Laboratório de Ecologia, Manejo e Conservação de Fauna Silvestre, Departamento de Ciências Florestais, Escola Superior de Agricultura ‘Luiz de Queiroz’, Universidade de São Paulo, Piracicaba, SP, Brazil 2 Laboratório de Ecologia Espacial e Conservação, Departamento de Ecologia, Universidade Estadual Paulista ‘Júlio de Mesquita Filho’, Rio Claro, SP, Brazil 3 Área de Relevante Interesse Ecológico ‘Matão de Cosmópolis’, Instituto Chico Mendes de Conservação da Biodiversidade, Ministério do Meio Ambiente, Campinas, SP, Brazil Keywords biodiversity; ecological functions; conservation planning; Brazil; patch size; mammal assemblages; fragmented habitat; functional diversity. Correspondence Marcelo Magioli, Departamento de Ciências Florestais, Escola Superior de Agricultura ‘Luiz de Queiroz’, Universidade de São Paulo, Av. Pádua Dias, 11, Bairro Agronomia, CEP 13418-900, Piracicaba, SP, Brazil. Email: marcelo.magioli@gmail.com Editor: Res Altwegg Associate Editor: Nick Isaac Received 24 September 2014; accepted 16 February 2015 doi:10.1111/acv.12201 Abstract Patch size affects abundance and diversity of mammal species, but there is little information on threshold regarding this relationship or relating it to functional diversity. Therefore, we aimed with this study to (1) evaluate if the functional diversity of medium- and large-sized mammal assemblages can be explained by patch size; (2) if this relationship is positive, evaluate whether it is linear (neutral hypothesis) or has critical thresholds (threshold hypothesis); (3) propose specific conservation strategies for each situation. We used primary and secondary mammal database for different forest remnants sizes in the Brazilian Atlantic Forest biome. We calculated functional diversity (FD), using an ecologically meaningful set of traits: body mass and locomotion form, behavioral and dietary traits, and environmental sensitivity of species. We compared linear models with threshold models using Akaike information criterion (AIC). FD values increased with patch size, possibly associated with the high complexity and heterogeneity of larger areas. The threshold model better explained the pattern between FD values and patch sizes than the linear one (delta AIC = 35.8), confirming our threshold hypothesis. Two thresholds (at 60 and 2050 ha) were identified. Our results high- light the need of compliance with the Brazilian Forest Code for assemblages in fragments lower than 60 ha. For assemblages in fragments between the two thresholds whose FD values significantly increased with patch size, we recommend improvements in conservation planning. For assemblages in fragments bigger than 2050 ha, we suggest the establishment of new protected areas, or at least, the maintenance of the existent ones. Our approach can serve as the basis for analysis with other taxonomic groups and ecosystems, increasing chances of maintaining faunal ecological functions, and improving species conservation. Introduction Ecological thresholds between response and explanatory variables have been pursued in recent years (Groffman et al., 2006), particularly to support conservation and bio- logical restoration plans (Briske, Fuhlendorf & Smeins, 2006). By definition, thresholds are abrupt changes from one ecological condition to another, which are driven by small additional changes in parameters in an already com- promised situation (Bennett & Radford, 2003). Thresholds may reflect species sensitivity to processes such as habitat loss or fragmentation (Lindenmayer & Luck, 2005). Thresh- olds aiming at species conservation have been identified for birds (Radford, Bennett & Cheers, 2005; Martensen et al., 2012), plants, frogs, fungi (Drinnan, 2005) and small mammals (Pardini et al., 2010; Estavillo, Pardini & Rocha, 2013). Although acting directly and indirectly on structuring and maintenance of natural environments through funda- mental ecosystem services such as seed dispersal and preda- tion (Jorge et al., 2013), there are no records about the existence of thresholds for the response of medium- and large-sized mammals to landscape structure. It is known that patch size affects abundance and diver- sity of mammal species (August, 1983; Pardini et al., 2005; Michalski & Peres, 2007), but there is little information on thresholds regarding this relationship; for example, what is the limit of habitat loss or patch size reduction that a species can tolerate before becoming locally extinct (Lindenmayer bs_bs_banner Animal Conservation. Print ISSN 1367-9430 Animal Conservation •• (2015) ••–•• © 2015 The Zoological Society of London 1