- Top of page
One of the fundamental goals of ecology, as well as a cause of intense debate during the last decades, is to elucidate the factors, processes and mechanisms that drive β-diversity (Tuomisto et al. 1995, 2003; Tuomisto 2010). This debate has been partly confounded by the diversity of concepts, methods and measures of β-diversity that have been used in the last few decades (Anderson et al. 2010). There are two fundamental concepts of β-diversity: turnover and variation (Anderson et al. 2010). In this study we focus on β-diversity defined as a variation in community composition among sample units within a spatial extent, without any reference to a particular gradient or direction. Regardless of the concept used, there are two main types of measure of β-diversity: (1) classical metrics that consider only the number of species (presence–absence data) and provide a direct link between local (α) and regional (γ) diversity (Blackburn & Gaston 1996; Whittaker 1960; Whittaker 1972;); and (2) multivariate measures, based on pair-wise resemblances among sample units, which take into account species composition (abundance, biomass or cover). Since these two types of measure can yield very different results, it is essential to employ both to address different and complementary aspects of β-diversity. Clearly defining and measuring β-diversity as well as understanding the factors that affect it is important because it allows us to help evaluate management and conservation strategies at the local, landscape and regional levels (Halffter 1998; Nekola & White 1999).
Different theories about the origin and maintenance of species diversity, such as the niche and the neutral theories, postulate contrasting underlying causes of variation in community composition – β-diversity (Condit et al. 2002; Tuomisto et al. 2003). The niche theory (Hutchinson 1957; Leibold 2008) states that β-diversity is positively associated with environmental heterogeneity and that some species consistently perform better than others under certain environmental conditions. Consequently, species abundance is an indicator of how suitable the environmental conditions are for the species to grow and reproduce. Conversely, the neutral theory establishes that all species are competitively equal and able to grow equally well under a range of environmental conditions, and that species composition fluctuates because of dispersal limitation; if species assemblages are assumed to be spatially structured, it follows that sites that are close together will be more similar in composition than sites that are far apart (Hubbell 2001; Condit et al. 2002). Therefore the challenge is to understand the relative contribution of these two processes (niche partitioning vs dispersal limitation) to ecosystems β-diversity.
Landscape structure, measured as a function of the size, shape, similarity, contrast and other metrics of the geometry of patches (Gustafson 1998; FRAGSTATS, University of Massachusetts, Amherst, MA, USA), can influence β-diversity through multiple processes such as deforestation, fragmentation and the associated reduction of habitat, change in the amount of habitat edge, and increase in fragment isolation (Fahrig 2003). Most studies on the effects of landscape patterns on plant communities, however, focus on local number of species (α-diversity) (Hernández-Stefanoni et al. 2011).
Secondary forest succession can also influence β-diversity, through changes both in species richness – classical metrics – and in species composition – multivariate measures – (see reviews of Guariguata & Ostertag 2001; Chazdon et al. 2007; Chazdon 2008; Quesada et al. 2009 for tropical wet and dry forests). However, very few studies have explicitly explored successional changes in β-diversity or examined the influence of successional age on β-diversity of tropical forests (Chao et al. 2005, 2006).
Moreover, the degree to which environmental factors affect variation in community composition depends on processes occurring at different spatial scales (Shmida & Wilson 1985; Levin 1992; McGill 2010). For example, local species composition is influenced by the species’ ability to compete for limited resources like water or nutrients (Tilman 1994). In turn, at the landscape level (mosaic of different patches), species composition is influenced by dispersal and the existence of different habitats that allow the co-existence of species with different niches (Bell et al. 2000). Therefore, the study of β-diversity should focus not only on the relative effects of different factors, but it should also evaluate how their relative influence changes across different spatial scales. Here, we examine one component of the scale, namely the grain (i.e. the size of the observation unit), using two different sizes: 200 m2 and 1 km2, henceforth referred to as local and landscape grain, respectively.
Evaluating the relative influence of factors, processes and mechanisms on β-diversity at different spatial scales is especially relevant for tropical dry forests (TDFs) because of their particularly high levels of endemism, coupled with high rates of deforestation, degradation and fragmentation, and low levels of protection (Miles et al. 2006; Pennington et al. 2009; Portillo-Quintero & Sánchez-Azofeifa 2010). The main objectives of this study were (1) to assess the effect of variation in spatial grain and forest successional age on the magnitude of β-diversity, and (2) to evaluate the relative importance of environmental heterogeneity, forest successional age, landscape structure and spatial structure of sampling sites on variation in TDF species composition at two contrasting spatial grains (local and landscape). Through this procedure we were able to indirectly gauge the relative importance of niche segregation and of dispersal limitation on β-diversity in a TDF.
Several theoretical principles guided us in making four predictions to be examined in this investigation. First, we predicted that the magnitude of β-diversity would be higher at the local grain size compared to the landscape size – as long as γ-diversity remains constant (P1). This prediction is based on the widely accepted notion of β-diversity as the γ-diversity/α-diversity ratio, i.e. the division of the total number of species in a landscape or region by the mean number of species in a locality or plot (Magurran 2004). Second, we predicted that the correlation between β-diversity and the explanatory variables would be stronger at the landscape grain than that at the local grain (P2). This prediction relies on the tenet that grain size produces a strong stochastic error component on the analysis of β-diversity: the smaller the grain, the larger the noise or stochastic error (Nekola & White 1999). Third, we predicted significant differences in species composition among successional age classes, and a decline in β-diversity as successional age increased (P3). This prediction is based on the well-established changes in species composition, and increase in α-diversity during neotropical forest succession (Guariguata & Ostertag 2001; Chazdon 2008; Quesada et al. 2009), which implies a corresponding decline in classical measures of β-diversity. Fourth, we predicted that the relative importance of environmental heterogeneity and landscape structure (which are indirect indicators of niche partitioning) vs spatial structure of samples (an indirect indicator of dispersal limitation) would depend on the spatial grain: at the local grain, the influence of spatial structure of samples would dominate, whereas at the landscape grain environmental heterogeneity and landscape structure would have a greater influence on beta-diversity (P4). This prediction is based on the recently observed scale dependence of the relative importance of niche partitioning and dispersal limitation (Bell 2005; Cottenie 2005); at the local scale, environmental gradients tend to be shorter thus providing less room for the packing of species with different niches, thereby increasing the relative importance of dispersal limitation and stochastic variation (He & Legendre 2002; Duque et al. 2009). Conversely, moving to a larger spatial grain often results in an increase in environmental heterogeneity and a concomitant increase in among-site species similarity (reduction in β-diversity), thereby enhancing the importance of niche partitioning relative to dispersal limitation (Bell 2005; Laliberté et al. 2009; Borcard et al. 2011).