Habitat Amount Hypothesis

Habitat Amount Hypothesis (abbreviated HAH) is a theoretical framework in landscape ecology and conservation biology formulated by the Canadian ecologist Lenore Fahrig in 2013 (Journal of Biogeography) that challenges the traditional patch-centered paradigm of island biogeography and metapopulation biology in terrestrial landscapes. Whereas classical fragmentation theory treats human-modified landscapes as collections of discrete habitat "islands" embedded in a hostile non-habitat matrix—predicting that species density in a sample site is governed jointly by patch size (the area A_patch of the contiguous fragment containing the sample plot) and patch isolation (the distance d_isol from that patch to neighboring patches)—Fahrig proposed that both variables can be replaced by a single, unified landscape predictor: the total amount of habitat (A_local) within an ecologically scaled circular neighborhood ("local landscape") surrounding the sample plot. According to the habitat amount hypothesis, apparent positive effects of patch size and negative effects of patch isolation in empirical field studies are largely statistical artifacts of the sample area effect, because larger and less isolated patches are almost always embedded within local landscapes that contain a larger total area of habitat.

The Classical Island-Biogeographic Patch Paradigm

Since the 1970s, applications of Robert H. MacArthur and Edward O. Wilson's (1967) equilibrium theory of island biogeography and Ilkka Hanski's (1998) spatially realistic metapopulation theory to terrestrial habitat fragmentation have rested on two core assumptions:

1. Discrete Patch Boundaries: Contiguous habitat patches (such as forest fragments surrounded by agricultural fields) act as bounded population units whose local extinction rate depends on the area of the individual patch (A_patch).
2. Inter-Patch Isolation: Re-colonization of an empty or declining patch depends on the Euclidean or edge-to-edge distance (d_isol) between that patch and the nearest occupied source patch(es).

Under this traditional paradigm, subdividing a fixed total habitat area A_local into several smaller patches (habitat fragmentation per se, holding total habitat amount constant) is predicted to reduce local species density (S_plot) by shrinking individual patch areas (A_patch) and increasing edge-to-interior ratios.

Fahrig's (2013) Sample Area Effect and the Local Landscape

Fahrig (2013) argued that for most mobile organisms inhabiting terrestrial landscapes, patch boundaries are permeable and individuals move through the landscape matrix among multiple habitat elements within their daily foraging or natal dispersal radius R_scale (the scale of effect). Consequently, the effective pool of conspecifics and heterospecifics capable of colonizing a standardized sample plot of fixed area a₀ (for example, a 1-hectare forest plot or a 50-meter point-count circle) is determined not by the arbitrary boundaries of the contiguous patch in which the plot sits, but by the total habitat area A_local summed across all patches within the circular local landscape of radius R_scale centered on the plot:

S_plot(a₀) = f( A_local ) , where A_local = ∑_k Area( Patch_k ∩ Circle(R_scale) )

Fahrig termed this mechanism the sample area effect: a standardized sample plot of area a₀ draws individuals from the surrounding local habitat pool A_local. By the fundamental species–area relationship (S_pool = c A_local^z) and the more-individuals hypothesis, a local landscape with greater total habitat A_local supports a larger total community abundance and a richer local species pool S_pool, increasing the expected number of species S_plot captured inside the fixed sample plot a₀.

To distinguish the Habitat Amount Hypothesis from the classical patch-size/isolation paradigm without confounding total habitat loss with spatial configuration, Fahrig (2013) specified a strict experimental design using equal-sized sample plots (a₀ = constant; never whole-patch censuses, which trivially confound sampling effort with patch size) across independent gradients of patch size (A_patch) and local habitat amount (A_local):

1. Dominance of Local Habitat Amount (A_local): Across equal-sized sample sites a₀, species richness S_plot increases monotonically with the total habitat amount A_local in the local landscape, even when the size of the sampled patch A_patch is held constant.
2. Null Effect of Patch Size (A_patch) Given A_local: When total habitat amount A_local in the surrounding local landscape is held constant, species richness S_plot in an equal-sized sample plot is independent of the size A_patch of the individual patch containing the plot. For example, a 1-ha sample plot inside a single 20-ha forest patch surrounded by 80 ha of non-habitat (A_local = 20 ha, A_patch = 20 ha) should contain the same number of species as a 1-ha sample plot inside a 2-ha forest patch whose local landscape contains nine additional 2-ha forest patches (A_local = 20 ha, A_patch = 2 ha).
3. Null Effect of Patch Isolation (d_isol) Given A_local: Because classical isolation metrics (such as nearest-neighbor distance or proximity indices) are mathematical inverses of surrounding habitat amount A_local, patch isolation has no independent negative effect on S_plot once A_local is statistically controlled.
4. Scale-of-Effect Scaling with Organism Mobility: The characteristic radius R_scale at which A_local best predicts S_plot scales positively with the dispersal mobility and home-range size of the focal taxonomic group (ranging from 100–500 m for flightless litter beetles and bryophytes to 2–10 km for forest birds and medium-sized mammals).

Quantitative Matrix of Predictions: Habitat Amount Hypothesis vs. Classical Patch-Fragmentation Models

Individual-Based Dispersal Simulations (Jackson and Fahrig, 2012–2016)

Using spatially explicit individual-based population simulations on fractal landscapes, HB Jackson and Lenore Fahrig (2012, 2014) and Fahrig (2013) demonstrated that the sample area effect emerges whenever organisms can cross small gaps of matrix without 100% immediate mortality. Only when two extreme biological conditions hold simultaneously—(1) per-step matrix mortality approaches 100% and (2) species have strict interior area requirements with severe edge avoidance—does patch size A_patch exert an independent positive effect beyond A_local.

Meta-Analyses and the Hanski–Haddad–Fletcher Counter-Critiques

The publication of the Habitat Amount Hypothesis sparked one of the most vigorous empirical and theoretical debates in modern conservation ecology:

• Critiques from Metapopulation and Fragmentation Ecologists (Saura, 2021):
  1. Ilkka Hanski argued that while A_local approximates metapopulation capacity in highly covered landscapes, below the extinction threshold (~10%–20% habitat cover) spatial clustering of habitat into larger or aggregated patches is essential to maintain metapopulation persistence.
  2. Nick M. Haddad et al. (2015, 2017) analyzed long-term controlled fragmentation experiments (including the Savannah River corridor project, the Biological Dynamics of Forest Fragments Project [BDFFP] in the Brazilian Amazon, and the Wog Wog eucalyptus experiment in Australia) and Robert J. Fletcher et al. argued that edge effects, microclimatic desiccation, and matrix hostility cause systematic species loss in small fragments even when A_local is held constant.
  3. Santiago Saura (2021) demonstrated mathematically using continuum percolation and spatial point-process models that the Habitat Amount Hypothesis holds exactly when inter-patch dispersal distance d_disp is larger than the mean inter-patch spacing, but breaks down for dispersal-limited interior specialists (d_disp ≪ patch spacing) and edge-sensitive taxa.
• Empirical Meta-Analyses (Martin, 2018; Watling et al., 2020):
  1. Charles A. Martin (2018) synthesized 28 independent field datasets testing HAH and found that while A_local was a strong positive predictor of plot richness, patch size A_patch retained a small positive residual effect in roughly half of studies (especially in hostile agricultural or aquatic matrices).
  2. James I. Watling et al. (2020) analyzed 35 global multi-taxa datasets (spanning birds, mammals, reptiles, amphibians, invertebrates, and vascular plants across >1,300 sample sites) specifically standardized by equal sample plot area a₀. Watling et al. found that total habitat amount A_local was a vastly stronger and more consistent predictor of species richness than patch size or configuration, supporting the primary prediction of the Habitat Amount Hypothesis across most terrestrial assemblages.

The Habitat Amount Hypothesis carries direct operational consequences for global biodiversity conservation and land-use policy:

• Priority of Stopping Total Habitat Loss: Because total habitat amount A_local governs the overwhelming majority of variance in local species richness, conservation policies must prioritize preventing net habitat loss and restoring total habitat area rather than expending scarce conservation funds purely on spatial configuration or rejecting small habitat patches.
• Conservation Value of Small Patches: Under the traditional patch-size paradigm, small forest patches, hedgerows, urban woodlots, and wetland remnants (<1–5 ha) are frequently dismissed as having negligible conservation value and cleared under environmental offsetting schemes. In contrast, the Habitat Amount Hypothesis—supported by Fahrig's (2017, 2020) syntheses of SLOSS datasets—demonstrates that every hectare of small patches contributes directly to A_local and often supports equal or higher cumulative landscape biodiversity than an equivalent area consolidated into a single large reserve.

Why must tests of the habitat amount hypothesis use equal-sized sample plots rather than whole-patch surveys?

If an ecologist censuses the entire area of a 100-ha patch and compares it to the entire area of a 2-ha patch, the 100-ha survey covers 50 times more sampling area than the 2-ha survey and will always record more species simply because of the species–area relationship (passive sampling effort). To test whether a 100-ha patch is intrinsically better habitat per unit area than a 2-ha patch, one must compare equal-sized sample plots (for example, one 1-ha plot inside the 100-ha patch vs. one 1-ha plot inside the 2-ha patch) across landscapes with controlled total habitat amount A_local.

What is the "scale of effect" in the habitat amount hypothesis, and how is it determined?

The scale of effect (R_scale) is the radial distance around a focal sample site over which total surrounding habitat amount A_local exerts its strongest statistical effect on local species richness S_plot. Empirically, researchers measure habitat area within nested concentric buffers around every sample plot (e.g., radii of 100 m, 250 m, 500 m, 1 km, 2 km, and 5 km), regress S_plot against A_local(R) at each radius, and identify R_scale as the radius yielding the maximum log-likelihood, R², or minimum AIC.

Under what biological conditions does the habitat amount hypothesis break down?

As formalized by Fahrig (2013) and Saura (2021), patch size A_patch exerts an additional positive effect independent of local habitat amount A_local under two specific conditions: (1) when the surrounding non-habitat matrix is an impenetrable barrier with near-total mortality (such as true oceanic islands for terrestrial flightless vertebrates, where individuals cannot move among patches within the local landscape), or (2) when focal species are interior-obligate specialists experiencing high abiotic or biotic edge mortality across a wide edge buffer zone d_edge that eliminates core habitat inside small patches.

1. Fahrig, L. (2013). "Rethinking patch size and isolation effects: The habitat amount hypothesis." Journal of Biogeography, 40(9), 1649–1663. https://doi.org/10.1111/jbi.12130
2. Haddad, N. M., Gonzalez, A., Brudvig, L. A., Burt, M. A., Levey, D. J., & Damschen, E. I. (2017). "Experimental evidence does not support the habitat amount hypothesis." Ecography, 40(1), 48–55. https://doi.org/10.1111/ecog.02535
3. Martin, C. A. (2018). "An early synthesis of the habitat amount hypothesis." Landscape Ecology, 33(11), 1831–1835. https://doi.org/10.1007/s10980-018-0716-y
4. Watling, J. I., Arroyo-Rodríguez, V., Pfeifer, M., Baeten, L., Banks-Leite, C., Cisneros, L. M., Fang, R., Hamel, A. C., Lachat, T., Leal, I. R., Lens, L., Possingham, H. P., Raheem, D. C., Ribeiro, D. B., Slade, E. M., Urbina-Cardona, J. N., Wood, E. M., & Fahrig, L. (2020). "Support for the habitat amount hypothesis from a global synthesis of species density studies." Ecology Letters, 23(4), 674–681. https://doi.org/10.1111/ele.13471
5. Saura, S. (2021). "The Habitat Amount Hypothesis implies negative effects of fragmentation on short-dispersed species." Journal of Biogeography, 48(1), 11–22. https://doi.org/10.1111/jbi.13958