Ecological Assembly Rules

Ecological assembly rules are quantitative principles and mechanistic constraints that govern how local biological communities are non-randomly selected and constructed from a regional species pool across an ecosystem (Diamond, 1975; Keddy, 1992; Weiher & Keddy, 1995). Coined by American biogeographer and physiologist Jared M. Diamond in 1975 from his survey of landbird distributions across the Bismarck Archipelago of Papua New Guinea and broadened into a functional-trait filtering framework by Canadian plant ecologist Paul A. Keddy (1992), assembly-rule theory seeks to explain why only a restricted, non-random subset of the species capable of dispersing to a habitat patch coexist there—and why species co-occurrence matrices, functional trait distributions, and community phylogenies exhibit predictable signatures of trait convergence (underdispersion) or limiting similarity (overdispersion).

Within community ecology and the study of the biosphere, assembly rules bridge the gap between regional macroecology (island biogeography and the mid-domain effect) and local coexistence mechanisms (Lotka–Volterra competition, resource partitioning, and historical priority effects).

In his foundational 1975 chapter "Assembly of Species Communities," Jared M. Diamond analyzed the presence–absence matrix of 142 breeding landbird species across 50 islands of the Bismarck Archipelago northeast of New Guinea. Observing that ecologically similar species within the same foraging guild (such as congeneric cuckoo-doves Macropygia mackinlayi and Macropygia nigrirostris, or gleaning honeyeaters Myzomela pammelaena and Myzomela sclateri) rarely co-occurred on the same island even when both species inhabited nearby islands of comparable area, Diamond formulated assembly rules governed by diffuse interspecific competitive exclusion:

1. Forbidden Species Combinations: Out of all mathematically possible subsets of species drawn from a guild of S regional species, only certain "permissible combinations" exist in nature; other pairs or trios are unstable and mutually exclusive.
2. Checkerboard Distributions: Two closely related or functionally similar competitors occupy mutually exclusive islands across an archipelago—forming a checkerboard pattern of 1s and 0s in the species-by-island presence–absence matrix—such that any island harbors Species A or Species B, but never both together.
3. Incidence Functions and Supertramps: Plotting the proportion of islands occupied (J) against island species richness (S) reveals distinct ecological strategies: high-S specialists restricted to species-rich mainland-like islands, and fugitive "supertramp" generalists (such as Macropygia mackinlayi) with high propagule pressure that colonize small, disturbed, species-poor islands but are competitively excluded from species-rich islands.

Diamond's 1975 assembly rules ignited a major methodological debate in 20th-century ecology. In 1979, Edward F. Connor and Daniel Simberloff challenged Diamond's interpretation in "The Assembly of Species Communities: Chance or Competition?" They argued that in any sparse presence–absence matrix where species differ in how many islands they occupy (row totals r_i) and islands differ in total species richness (column totals c_j), many pairs of species fail to co-occur by chance without interspecific competition.

To test whether observed species co-occurrences depart from random colonization, Connor and Simberloff (1979) and later Lewi Stone and Alan Roberts (1990) and Nicholas J. Gotelli (2000) developed Monte Carlo binary matrix permutation algorithms (such as the fixed–fixed SIM9 sequential swap algorithm, which preserves both empirical species occurrence frequencies r_i and island richness totals c_j). Across thousands of randomized null matrices, ecologists compute the Stone and Roberts (1990) Checkerboard Score (C-score):

• Pairwise Checkerboard Units (CU_ij): CU_ij = (r_i − S_ij) (r_j − S_ij)
• Community Mean C-Score: C = [ 2 / (S (S − 1)) ] ∑ (r_i − S_ij) (r_j − S_ij)
• Standardized Effect Size (SES): SES = (C_obs − C_null) / σ_null

where r_i and r_j are the total number of sites occupied by species i and j, S_ij is the number of sites where both species co-occur, S is the number of species in the matrix, and C_null and σ_null are the mean and standard deviation of the C-score across 1,000 randomized null matrices. When SES exceeds +1.96 (C_obs significantly exceeds C_null), the community exhibits non-random species segregation consistent with competitive exclusion or habitat sorting; across 96 published presence–absence matrices re-analyzed by Gotelli and McCabe (2002), animal and plant assemblages exhibited a mean C-score SES of +2.90, confirming widespread non-random community structure.

In 1992, Paul A. Keddy reformulated assembly rules around functional traits (measurable morphological, physiological, or phenological attributes such as specific leaf area, wood density, seed mass, root depth, or beak size) rather than taxonomic species names. Together with Evan Weiher (Weiher & Keddy, 1995) and David D. Ackerly (Cornwell & Ackerly, 2009), Keddy modeled community assembly as a sequential hierarchical sieve of three filters acting on the regional species pool:

Opposing Trait Signatures: Environmental Filtering vs. Limiting Similarity

As synthesized by Evan Weiher and Paul A. Keddy (1995) and tested in California chaparral woody plants by William K. Cornwell and David D. Ackerly (2009), the abiotic filter and the biotic competition filter leave opposite statistical signatures on local trait distributions:

1. Abiotic Environmental Filtering Drives Trait Convergence (Underdispersion): Harsh abiotic conditions—such as serpentine soils, waterlogged anoxic wetlands, or alpine frost zones—eliminate all regional species lacking the required physiological adaptations. Consequently, co-occurring species inside a stressful patch exhibit a narrower range and variance of functional traits than random draws from the regional pool.
2. Biotic Limiting Similarity Drives Trait Divergence (Overdispersion): Within benign, resource-rich neighborhoods where plants or animals compete for light, soil nitrate (NO₃⁻), orthophosphate (PO₄³⁻), or prey size classes, Robert MacArthur and Richard Levins's (1967) principle of limiting similarity predicts that species with nearly identical functional traits cannot stably coexist. Competitive exclusion removes intermediate phenotypes, leaving surviving neighbors more evenly spaced (overdispersed) along functional trait axes (such as rooting depth or bill depth in Darwin's finches Geospiza) than expected by chance.

In 2002, Campbell O. Webb, David D. Ackerly, Mark A. McPeek, and Michael J. Donoghue integrated assembly rules with molecular phylogenetics. When functional traits are phylogenetically conserved across evolutionary history, abiotic environmental filtering produces phylogenetic clustering (Net Relatedness Index NRI > 0), whereas competitive exclusion (or host-specific Janzen–Connell pathogen predation) produces phylogenetic overdispersion (Nearest Taxon Index NTI < 0)—as seen in Quercus oaks in Florida sandhills (Cavender-Bares et al., 2004). However, under Peter Chesson's modern coexistence theory (HilleRisLambers et al., 2012), competitive exclusion can also drive trait clustering whenever a single trait (such as tall canopy height) confers an overarching competitive hierarchy rather than niche differentiation across alternative stable states.

1. Diamond, J. M. (1975). "Assembly of species communities." In M. L. Cody & J. M. Diamond (Eds.), Ecology and Evolution of Communities (pp. 342–444). Harvard University Press. https://doi.org/10.2307/1936581
2. Connor, E. F., & Simberloff, D. (1979). "The assembly of species communities: chance or competition?" Ecology, 60(6), 1132–1140. https://doi.org/10.2307/1936961
3. Keddy, P. A. (1992). "Assembly and response rules: two goals for predictive community ecology." Journal of Vegetation Science, 3(2), 157–164. https://doi.org/10.2307/3235676
4. Webb, C. O., Ackerly, D. D., McPeek, M. A., & Donoghue, M. J. (2002). "Phylogenies and community ecology." Annual Review of Ecology and Systematics, 33, 475–505. https://doi.org/10.1146/annurev.ecolsys.33.010802.150448