Baselga lab

Multi-hierarchical macroecology

Aims

Understanding the causes of macroecological patterns has proven difficult, despite major advances derived from empirical studies of large-scale statistical patterns of species richness, composition and abundance (e.g., Brown 1995; Nekola & White 1999; Willig, Kaufman & Stevens 2003). Several “unified” theories have been developed that explain regularities in these macroecological distributions, although they emphasize different underlying processes regarding the role of niche-derived and neutral factors (McGill 2010). Despite their fundamental differences, various models – including the Neutral Theory of Biodiversity (Hubbell 2001; Rosindell, Hubbell & Etienne 2011) and the Ecological Niche Theory (Hutchinson 1957; Leibold 1995; Peterson et al. 2011) – are consistent with widely observed macroecological patterns such as the decline of community similarity with spatial distance (i.e. distance-decay of similarity, Nekola & White 1999), the species abundance–range size correlations (McGill et al. 2007), or the species–area curves (Drakare, Lennon & Hillebrand 2006). Because coinciding patterns are predicted from competing mechanisms, it has been shown that discerning the effects of neutral and non-neutral ecological processes on biodiversity is usually difficult.

The integrated assessment of macroecological patterns at various levels of genotypes, genealogies and species (i.e. multi-hierarchical macroecology, see Baselga et al. 2013; Baselga, Gómez-Rodríguez & Vogler 2015) provides a novel avenue to disentangle the competing potential drivers of biodiversity patterns. In the figure below, it is shown how the lineage branching with evolutionary time could be linked to macroecological patterns at multiple hierarchical levels by assessing the distribution ranges of lineages of different ages, i.e. between haplotypes (level 4, young lineages) to species (level 1, old lineages).

Multi-hierarchical macroecology

In the figure above we can see that if lineages' spatial distributions are neutral (i.e. controlled by limited stochastic dispersal) the ranges' size (represented by the coloured ovals on the top) just depends on lineage age, and the ranges of younger lineages are uniformly distributed within the ranges of older lineages. Thus, only under neutral evolutionary and ecological dynamics, lineages' distributions form a spatial fractal mirroring the fractal branching of lineages through time.

Roots

This approach has antecedents in studies assessing the species–genetic diversity correlation (SGDC, Vellend 2003), suggesting that species richness of a local assemblage is correlated with the genetic diversity within each of the local species because both patterns may be driven by a single process (Vellend et al. 2014). The SGDC was also observed in analyses of beta diversity that show a regular decay of assemblage similarity with geographic distance at both species and genetic levels (Sei, Lang & Berg 2009; Odat et al. 2010; Papadopoulou et al. 2011). This can be done by measuring similarity at both genetic and species levels (i.e. the proportion of haplotypes and species shared by two assemblages, see figure below).

Measuring beta at species and genetic levels

In the figure above it is shown (A) that dissimilarity is different at the species (2 species shared out of 4) and the haplotype level (2 haplotypes shared out of 12). If we measure this at multiple spatial distances, we can build distance-decay curves for both levels (B).

Although opening promising ways to investigate the processes underlying biodiversity, it has been shown that the SGDC alone does not suffice to separate neutral from non-neutral drivers, as both niche and neutral processes may result in the SGDC (Vellend & Geber 2005; Vellend et al. 2014), and even neutral processes alone can generate correlations of opposite sign between species and genetic diversity (Laroche et al. 2015).

A novel approach

Multi-hierarchical macroecology is related to the SGDC, but investigates the spatial patterns of variation of entire assemblages at multiple hierarchical levels between haplotypes and species, rather than the correlation between single species genetic diversity with species diversity. The levels below species can be studied by grouping genetic variants into increasingly more inclusive genealogical entities based on the relationships of haplotypes. In the case of non-recombining mitochondrial genomes, these haplotype groups reflect different depths of time of the gene genealogy extending back to an ancestor near the species origin. As gene copies acquire mutations during lineage history, they also are subject to dispersal, which extends their spatial ranges and builds up spatial patterns of assemblage variation. Macroecological analyses can therefore be extended to lower hierarchical levels in analogy to the analysis of variation in species richness and species composition in classical macroecology. This extension to multiple hierarchical levels takes advantage of novel patterns emerging across levels, which might help to discern between neutral and non-neutral controls of assemblage variation (Baselga et al. 2013). For example, when assemblage similarity is measured at multiple levels, neutral dynamics predict that the distance decay curves have similar slopes but different intercepts that vary regularly from level to level (in a fractal way, detected as a tight log-log correlation between intercepts and hierarchical levels, see figure below from Baselga, Gómez-Rodríguez & Vogler 2015). This fractal pattern across hierarchical levels is not predicted if niche processes control species distributions.

Fractal patterns emerging across levels

Central hypothesis and unique predictions

The central hypothesis of multi-hierarchical macroecology is that the effect of neutral processes (including neutral mutation, dispersal limitation, birth and death of lineages) is uniform across hierarchical levels, while the effect of non-neutral processes differs among levels. In particular, Ecological Niche Theory (Hutchinson 1957; Leibold 1995; Soberón & Nakamura 2009) proposes that species distributions are constrained by abiotic (e.g., climate) and biotic factors (e.g., interactions), via their effects on the fitness of individuals and the viability of populations. In contrast, there is no suggestion that specific mitochondrial DNA (mtDNA) haplotypes are linked to a particular niche (i.e. different from haplotype niches of conspecifics). The evolution of mtDNA has been considered to be predominantly controlled by neutral processes (Slatkin 1985; Avise 1994). Some recent work proposed that mtDNA is controlled by selective processes (Bazin, Glemin & Galtier 2006), but this did not overturn the view that mtDNA variation is predominantly neutral (e.g., Mulligan, Kitchen & Miyamoto 2006; Albu et al. 2008; Nei, Suzuki & Nozawa 2010; Karl et al. 2012). In addition, evolutionary neutrality can be empirically tested in each specific study system by assessing how DNA variation is distributed across common and rare haplotypes (Tajima 1989). This establishes the basis of our predictive framework: neutral molecular evolution of mtDNA is evidence for the absence of selective regimes that would determine the spatial distribution of genetic variation. Hence, the spatial range of mtDNA haplotypes within a species range is likely to be controlled by neutral processes only (Slatkin 1985; Avise 1994). Consequently, the variation in haplotype composition of assemblages provides a benchmark of ecological neutrality against which to compare the variation of assemblages at the species level.

This leads to the major prediction of multi-hierarchical macroecology that, under neutral dynamics, the variation of composition of entire assemblages at and below the species level is self-similar at any temporal and spatial scale (Baselga et al. 2013). Self-similarity, in turn, should be eroded under the effect of niche processes at the species level as non-neutral processes affect spatial ranges of lineages differently at various levels. Therefore, assessing macroecological patterns at multiple hierarchical levels (between genes and species) produces unique predictions allowing to discern the relevance of neutral and non-neutral factors.

References

Inordinate fondness for beetles