Alpha, Beta, and Gamma Diversity Metrics

Alpha, beta, and gamma diversity metrics constitute the quantitative framework used in community ecology, biogeography, and conservation biology to measure and partition biodiversity across hierarchical spatial scales. Introduced by American plant ecologist Robert H. Whittaker in his 1960 monograph on the Siskiyou Mountains of Oregon and refined in his 1972 Taxon synthesis, this three-tier classification resolves a regional metacommunity or biogeographic province into three linked components: alpha diversity (α), the local within-habitat diversity of a single uniform sample unit or stand; gamma diversity (γ), the total regional diversity pooled across all habitats within a landscape; and beta diversity (β), the extent of compositional differentiation, species turnover, or community dissimilarity among local habitats along environmental or geographic gradients.

Beyond counting raw taxonomic species richness (S), modern diversity theory integrates relative species abundances (p_i), evolutionary branch lengths (phylogenetic diversity), and organismal trait hypervolumes (functional diversity). Furthermore, mathematical ecologist Lou Jost (2006, 2007) unified classical entropy and dominance indices under Hill numbers (the effective number of species, ^qD), restoring an intuitive linear scale that satisfies the replication principle required for rigorous conservation comparisons.

Species Richness, Rarefaction, and Chao1 Asymptotic Estimators

The simplest measure of alpha diversity is species richness (S), the count of species observed within a local quadrat or point count. However, empirical richness S_obs increases non-linearly with sampling effort (number of individuals N or plots A sampled) and systematically underestimates true community richness by missing rare species (Preston's veil line). To standardize richness across unequal sample sizes, Howard L. Sanders (1968) and Stuart H. Hurlbert (1971) developed individual-based rarefaction, calculating the expected number of species *E*[S_n] in a random subsample of n < N individuals drawn without replacement:

*E*[S_n] = ∑_i=1^S [1 − C(N − N_i, n) / C(N, n)]

To estimate the undetected asymptote of true richness S_true, statistician Anne Chao (1984) derived the non-parametric Chao1 estimator using Good–Turing cryptographic frequency theory, which infers the number of unrecorded species from the frequency of singletons (f₁, species represented by a single individual in the sample) and doubletons (f₂, species represented by two individuals):

Ŝ_Chao1 = S_obs + f₁² / (2*f*₂) (or S_obs + f₁(f₁ − 1) / [2(f₂ + 1)] for bias-corrected small samples)

Information-Theoretic Entropy, Dominance, and Hill Numbers

Because two communities with identical species richness S can differ vastly in evenness (whether abundances p_i = N_i/N are distributed equally or dominated by a single hyper-abundant taxon), ecologists employ abundance-weighted indices:

• Shannon–Wiener Entropy (H'): Derived from Claude Shannon's (1948) information theory, H' = −∑_i=1^S p_i ln p_i measures the uncertainty in predicting the species identity of a randomly drawn individual (ranging from 0 for a monoculture to ln S when all p_i = 1/S). Evelyn C. Pielou's Evenness (J') normalizes Shannon entropy by its theoretical maximum: J' = H' / ln S ∈ [0, 1].
• Simpson Concentration (λ or D): Introduced by Edward H. Simpson (1949), λ = ∑_i=1^S p_i² measures the probability that two individuals drawn at random belong to the same species (weighted heavily toward dominant species), commonly expressed as the Gini–Simpson index (1 − ∑ p_i²) or reciprocal Simpson index (1 / ∑ p_i²).
• Hill Numbers (Effective Number of Species, ^qD): Mark O. Hill (1973) and Lou Jost (2006) proved that raw Shannon entropy H' and Gini–Simpson 1 − λ are non-linear entropies rather than true diversities—violating the doubling property (combining two equally abundant, non-overlapping communities of S species should yield a diversity of 2*S*). Converting entropies into Hill numbers of order q yields the number of equally abundant species required to produce the observed index value:

^qD = ( ∑_i=1^S p_i^q )^{1 / (1 − q)}

At q = 0, ⁰D = S (species richness, weighting rare and common species equally); as q → 1, ¹D = exp(H') (exponential of Shannon entropy, weighting species strictly in proportion to their relative frequencies p_i); and at q = 2, ²D = 1 / ∑ p_i² (inverse Simpson index, weighting dominant species).

Multiplicative (Whittaker) versus Additive (Lande) Partitioning

Given a landscape of M sampled local sites with mean local diversity ᾱ and pooled regional diversity γ, Robert H. Whittaker (1960) defined multiplicative beta diversity:

β_W = γ / ᾱ ⇔ γ = ᾱ × β_W

where β_W ranges from 1 (when all M local sites have identical species composition) to M (when no two sites share any species) and represents the effective number of distinct community compositions in the region. Alternatively, Russell Lande (1996) popularized additive diversity partitioning (β_add = γ − ᾱ), which expresses ᾱ, β_add, and γ in the same units to evaluate the average number of species missing from a single local site, while Lou Jost (2007) demonstrated that multiplicative partitioning of Hill numbers (^qβ = ^qγ / ^qα) guarantees statistical independence between alpha and beta components.

Baselga's Decomposition: Spatial Turnover versus Nestedness

When comparing two sites with a shared species, b species unique to site 1, and c species unique to site 2, classic pairwise dissimilarity indices such as the Jaccard index (d_J = [b + c] / [a + b + c]) and Sørensen index (β_sor = [b + c] / [2*a* + b + c]) conflate two distinct ecological processes. Andrés Baselga (2010) algebraically decomposed Sørensen dissimilarity into spatial turnover (β_sim, Simpson dissimilarity)—the replacement of species in one site by different species in another along an environmental gradient—and nestedness-resultant dissimilarity (β_sne)—where a species-poor site is a strict subset of a species-rich site due to ordered extinction or colonization filters:

β_sor = β_sim + β_sne = [ min(b, c) / (a + min(b, c)) ] + [ (|*b* − *c*| / (2*a* + b + c)) × (a / (a + min(b, c))) ]

This distinction guides reserve design: high spatial turnover (β_sim) requires protecting multiple complementary sites across the landscape, whereas high nestedness (β_sne) prioritizes conserving the richest hotspot sites that harbor the full regional species pool.

Faith's Phylogenetic Diversity (PD) and Trait Hypervolumes

Taxonomic diversity metrics treat all species as ecologically and evolutionarily equivalent—counting two sibling species of Drosophila as equal in diversity to a lungfish and a sequoia. To incorporate evolutionary history, Daniel P. Faith (1992) defined Phylogenetic Diversity (PD) as the sum of the lengths of all branches on a rooted phylogenetic tree (in millions of years of divergence or nucleotide substitutions) that span the set of species present in a community. Communities containing evolutionarily distinct, deep-branching lineages (such as the tuatara Sphenodon punctatus or Amborella trichopoda) exhibit high PD and preserve greater evolutionary heritage.

Simultaneously, functional diversity quantifies the distribution of species within a multidimensional trait space defined by morpho-physio-phenological functional traits (such as specific leaf area, wood density, seed mass, or beak dimensions, related to the Hutchinsonian niche). Norman W. H. Mason and Sébastien Villéger (2008) partitioned functional diversity into three orthogonal metrics: Functional Richness (FRic), the convex hull volume occupied by the community in trait space; Functional Evenness (FEve), the regularity of abundance spacing along the minimum spanning tree in trait space; and Functional Divergence (FDiv), the degree to which biomass is concentrated near the outer margins of the trait hypervolume, complemented by C. R. Rao's Quadratic Entropy (Q = ∑∑ d_ijp_ip_j).

Why did Lou Jost argue that Shannon's H' is an entropy rather than a true diversity?

Consider a continent containing 16 equally abundant species (H' = ln 16 = 2.77; Gini–Simpson = 1 − 1/16 = 0.9375). If conservationists lose half of those species so only 8 equally abundant species remain, true biological diversity has dropped by 50%. However, Shannon entropy H' drops only from 2.77 to ln 8 = 2.08 (a 25% reduction), and Gini–Simpson drops from 0.9375 to 0.875 (only a 6.7% reduction), creating a misleading impression of stability in species-rich ecosystems. Converting H' and Simpson's index to Hill numbers (¹D = exp(H') and ²D = 1/∑*p*_i²) restores linearity: both drop from 16 to 8, accurately reflecting the 50% loss.

What is distance decay of similarity in beta diversity?

Introduced to biogeography by allometric and spatial ecologists including Tobler's First Law of Geography and formalized by Hanna Tuomisto and Kalle Ruokolainen, the distance decay of similarity describes the monotonic decrease in community compositional similarity (1 − β_sor) as geographic distance (d) between two sampled plots increases. Distance decay arises from two distinct mechanisms that can be separated via partial Mantel tests or canonical redundancy analysis: environmental deterministic filtering (spatially autocorrelated changes in soil, elevation, and climate) and neutral dispersal limitation (finite propagule dispersal kernels and ecological drift under Hubbell's Unified Neutral Theory).

How do functional redundancy and response diversity stabilize ecosystems?

When multiple species within a community occupy similar positions in functional trait space (such as several nitrogen-fixing pasture legumes or multiple grazing parrotfish species on a coral reef), the community exhibits functional redundancy. Thomas Elmqvist and colleagues (2003) demonstrated that ecosystem resilience depends on response diversity—variation in how functionally redundant species within the same guild respond to environmental disturbances (such as drought, thermal anomalies, or pathogens). When one species declines, a functionally similar guild member with a different thermal tolerance compensates, stabilizing ecosystem productivity through the portfolio (insurance) effect.

1. Whittaker, R. H. (1960). "Vegetation of the Siskiyou Mountains, Oregon and California." Ecological Monographs, 30(3), 279–338. doi:10.2307/1943563
2. Faith, D. P. (1992). "Conservation evaluation and phylogenetic diversity." Biological Conservation, 61(1), 1–10. doi:10.1016/0006-3207(92)91201-391201-3)
3. Jost, L. (2006). "Entropy and diversity." Oikos, 113(2), 363–375. doi:10.1111/j.2006.0030-1299.14714.x
4. Baselga, A. (2010). "Partitioning the turnover and nestedness components of beta diversity." Global Ecology and Biogeography, 19(1), 134–143. doi:10.1111/j.1466-8238.2009.00490.x