Source–Sink Dynamics

Source–Sink Dynamics—formulated in its canonical mathematical form in 1988 by the American ecologist H. Ronald Pulliam in The American Naturalist—is the foundational theoretical framework in spatial population ecology, metacommunity theory, and conservation biology describing how dispersal across heterogeneous landscapes links high-quality source habitats (where in-situ birth rates exceed death rates) with low-quality sink habitats (where local death rates exceed birth rates). Prior to Pulliam (1988) and earlier conceptual antecedents by Joseph Grinnell, William Lidicker (1975), and Beatrice Van Horne (1983), demographic models typically assumed that local population density within a habitat patch reflects its intrinsic carrying capacity and habitat suitability. Pulliam demonstrated that when organisms exhibit active habitat selection under preemptive breeding-site territoriality (an ideal despotic distribution), surplus individuals produced in small, productive source patches continually emigrate into larger, unproductive sink patches. Consequently, local population density can be decoupled from—or even negatively correlated with—intrinsic demographic productivity, allowing more than 90% of a species' regional population to reside in sink habitats that would go deterministically extinct if isolated from their sources.

Compartment-Specific Vital Rates and Net Dispersal Flux

Consider a regional population (or metapopulation) distributed across a landscape mosaic composed of discrete habitat compartments or patch types i = 1, 2, ..., m within an ecosystem. Over an annual reproductive and wintering cycle, let B_i denote the per-capita in-situ birth rate (specifically, the mean number of juvenile offspring produced per breeding adult in compartment i that survive until the pre-breeding census) and let D_i denote the per-capita annual adult death rate in compartment i (so that annual adult survival is P_A,i = 1 − D_i, while B_i = β_iP_J,i is the product of per-capita fecundity β_i and juvenile survival P_J,i).

In the absence of immigration or emigration, the intrinsic finite rate of population increase λ_i in habitat compartment i over one annual cycle is:

λ_i = 1 + B_i − D_i = P_A,i + β_iP_J,i

Pulliam (1988) defined source and sink habitats rigorously in terms of both their intrinsic demographic balance (λ_i) and their equilibrium net dispersal flux:

1. Source Habitat (B₁ > D₁, λ₁ > 1): A habitat compartment in which local in-situ reproduction exceeds local mortality over the long-term average. Because the compartment generates a net demographic surplus (B₁ − D₁ > 0), at demographic equilibrium the emigration rate E₁ exceeds the immigration rate I₁ (net exporter of individuals).
2. Sink Habitat (B₂ < D₂, λ₂ < 1): A habitat compartment in which local in-situ reproduction fails to replace local mortality (B₂ − D₂ < 0). If isolated from external immigrants (I₂ = 0), a sink subpopulation of initial size n₂(0) decays deterministically to extinction according to n₂(t) = n₂(0)λ₂^t. At demographic equilibrium, a persistent sink subpopulation is maintained because immigration I₂ exceeds emigration E₂ (net importer of individuals).

Contrast with Classical Levins and Hanski Metapopulations

The Pulliam source–sink framework differs fundamentally from the classical patch-occupancy metapopulation models of Richard Levins (1969) and the spatially realistic incidence function model of Ilkka Hanski (1994). In classical metapopulation theory, all habitat patches are capable of supporting positive local growth when colonized, and persistence arises from a stochastic balance between patch extinction (driven by demographic or environmental stochasticity) and recolonization of empty patches. In Pulliam's source–sink model, by contrast, the distinction between patches is deterministic and demographic: a true sink patch has a negative intrinsic growth rate (r₂ = lnλ₂ < 0) at all population densities and would fail to sustain a population even in the absence of stochastic fluctuations, yet it remains continuously occupied year after year due to steady annual rescue by immigrants from source patches.

Preemptive Territoriality and the Ideal Despotic Distribution

To explain why organisms ever enter or breed in a sink habitat where their expected fitness (λ₂ < 1) is less than unity, Pulliam (1988) combined demographic accounting with Stephen Fretwell and Henry Lucas's (1969) theory of ideal despotic habitat selection (preemptive breeding-site territoriality). Suppose a landscape contains two habitat types: a high-quality source habitat (compartment 1) containing a fixed maximum number of defendable breeding territories (or nest sites) K₁, and a lower-quality sink habitat (compartment 2) where breeding sites are abundant.

Within the source habitat, every adult that secures one of the K₁ territories achieves adult survival P_A and fecundity β₁, yielding finite growth rate λ₁ = P_A + β₁P_J > 1. Within the sink habitat, adults achieve survival P_A but lower reproductive output β₂ < β₁ (for example, due to higher nest predation or lower food availability), yielding a sub-replacement finite growth rate λ₂ = P_A + β₂P_J < 1.

What happens to an individual in the source habitat if all K₁ breeding territories are already occupied (n₁ > K₁)? Under preemptive territoriality, an unmated "floater" that remains in the overcrowded source patch without a territory has zero reproductive success (β_floater = 0), so its expected fitness in the saturated source is only P_A. By emigrating to an unoccupied territory in the sink habitat, however, that same individual achieves expected fitness λ₂ = P_A + β₂P_J > P_A. Therefore, even though the sink habitat cannot sustain a population on its own (λ₂ < 1), natural selection favors active emigration from a saturated source into a sink whenever breeding in the sink yields higher individual fitness than remaining a non-breeding floater in the source (λ₂ > P_A).

Coupled Population Dynamics and Equilibrium Partitioning

When individuals follow this optimal preemptive settlement rule—filling all K₁ source territories first and dispersing all surplus individuals (n₁ − K₁) into the sink—the coupled discrete-time recurrence equations for the total pre-breeding population N(t) = n₁(t) + n₂(t) are:

• When N(t) ≤ K₁ (source unsaturated): All individuals settle in the source (n₁(t) = N(t), n₂(t) = 0), and the population grows geometrically at rate λ₁: N(t + 1) = λ₁N(t).
• When N(t) > K₁ (source saturated at n₁(t) = K₁, with n₂(t) = N(t) − K₁ in the sink): N(t + 1) = λ₁K₁ + λ₂n₂(t).

Setting N(t + 1) = N(t) = K₁ + n̂₂ at demographic equilibrium (where n̂₁ = K₁) yields the Pulliam equilibrium equation:

K₁ + n̂₂ = λ₁K₁ + λ₂n̂₂

Solving for the equilibrium sink population size n̂₂ and the total regional population size N̂ = n̂₁ + n̂₂ gives:

n̂₂ = K₁(λ₁ − 1)/(1 − λ₂), N̂ = K₁(λ₁ − λ₂)/(1 − λ₂)

The equilibrium fraction of the regional population residing in the source (p̂₁) versus the sink (p̂₂) is therefore:

p̂₁ = n̂₁/N̂ = (1 − λ₂)/(λ₁ − λ₂), p̂₂ = n̂₂/N̂ = (λ₁ − 1)/(λ₁ − λ₂)

Why Over 90% of a Species Can Reside in Sink Habitats

Pulliam's equilibrium ratio n̂₂/n̂₁ = (λ₁ − 1)/(1 − λ₂) yields a striking insight for ecology and conservation:

• Suppose a songbird or small mammal population has a productive source habitat where each breeding pair produces an annual surplus (λ₁ = 1.50, so λ₁ − 1 = 0.50) and a large marginal sink habitat where annual reproduction nearly—but not quite—balances mortality (λ₂ = 0.95, so 1 − λ₂ = 0.05).
• Then the equilibrium sink-to-source population ratio is n̂₂/n̂₁ = 0.50/0.05 = 10.

In this scenario, 10 times as many individuals live in the sink habitat as in the source habitat (p̂₂ = 10/11 ≈ 90.9% of the total regional population in the sink, and only 9.1% in the source). If a wildlife manager or conservation agency surveys population abundance across the landscape and protects only the habitat type supporting the largest total population or highest local density (the sink), while allowing the small source habitat (K₁) to be destroyed by land conversion, the regional population N̂ will collapse deterministically to zero. This mathematically formalizes Beatrice Van Horne's (1983) decoupling principle: local habitat density alone is an unreliable—and frequently misleading—indicator of habitat quality without direct measurements of local survival P_A and fecundity β_i.

Watkinson and Sutherland's (1995) Pseudo-Sinks

In 1995, Andrew R. Watkinson and William J. Sutherland demonstrated that a habitat patch exhibiting net immigration (I₂ > E₂) and sub-replacement per-capita fitness at equilibrium (λ₂(n̂₂) < 1) is not necessarily a true Pulliam sink. Suppose patch 2 has a positive intrinsic growth rate at low density (λ₂(0) > 1) and its own positive local carrying capacity K₂ > 0, but continuous heavy immigration from an adjacent productive source pushes the local density in patch 2 above its local carrying capacity (n̂₂ > K₂).

Because density-dependent competition or nest predation depresses reproduction at high density (n̂₂ > K₂), the observed per-capita growth rate in patch 2 is less than unity (λ₂(n̂₂) < 1), making patch 2 appear identical to a true sink in short-term demographic censuses. However, if the source patch 1 were removed (stopping immigration), the population in patch 2 would not go extinct; instead, it would relax downward from n̂₂ to its self-sustaining local carrying capacity K₂ > 0. Watkinson and Sutherland termed such density-inflated habitats pseudo-sinks.

Ecological Traps and Maladaptive Habitat Selection

Whereas Pulliam's (1988) model assumes that organisms rank habitats accurately so that they always settle in the source first before spilling over into the sink, rapid anthropogenic landscape modification can decouple the environmental cues that animals use to assess habitat quality (such as vegetation structure, edge green-up, or polarized light reflection) from actual survival and reproductive success. First conceptualized by J. Edward Gates and Leslie W. Gysel in 1978 and integrated into source–sink mathematical models by H. Ronald Pulliam and Brent J. Danielson (1991), Richard Donovan, and Martin A. Schlaepfer, Michael C. Runge, and Paul W. Sherman (2002), an ecological trap is a sink habitat (λ₂ < 1) that organisms actively prefer over an available high-quality source habitat (λ₁ > 1).

Unlike a benign Pulliam sink—which absorbs only surplus individuals (N > K₁) and actually increases total regional population size (N̂ > K₁)—an attractive sink (ecological trap) drains breeding adults directly out of unsaturated source territories (n₁ < K₁). If the preference for the ecological trap and the landscape proportion of trap habitat exceed a persistence threshold, the composite growth rate of the regional population drops below unity (λ_regional < 1), driving both the source and the sink subpopulations to regional extinction.

Balanced Dispersal, Evolution of Niche Conservatism, and Rescue Effects

Robert D. Holt (1985, 1996) and Mark A. McPeek and Robert D. Holt (1992) extended source–sink theory to examine evolutionary consequences of asymmetric gene flow from sources to sinks:

1. Evolution of Niche Conservatism in Sinks: Because the gene pool inside a sink habitat is continually swamped by immigrants whose ancestors were selected for performance in the source habitat, local natural selection for adaptation to the stressful biotic or abiotic conditions of the sink is impeded by gene swamping and demographic asymmetry, providing a demographic explanation for niche conservatism at geographic range boundaries.
2. Temporal Source–Sink Inversions and the Spatial Insurance Hypothesis: As shown by Nicolas Mouquet, Michel Loreau, and spatial insurance hypothesis models, environmental fluctuations or bet-hedging dispersal strategies can cause patches to switch roles over time (a "source" during wet years becoming a "sink" during drought years), allowing regional persistence even when every patch acts as a temporary sink during unfavorable phases.

Source–sink dynamics have been empirically verified through long-term demographic mark–recapture, nest monitoring, and genetic assignment studies across diverse taxa:

• Forest Songbirds in Fragmented Agricultural Landscapes (Robinson et al. 1995): Across the midwestern United States, Scott K. Robinson and colleagues monitored over 5,000 nests of nine forest-breeding Neotropical migratory songbirds (including the wood thrush Hylocichla mustelina and ovenbird Seiurus aurocapilla). In small, fragmented woodlots (< 20% regional forest cover), high edge densities led to 50%–80% nest parasitism by brown-headed cowbirds (Molothrus ater) and elevated nest predation, depressing local λ₂ well below 1.0. These fragmented forest sinks remained populated only via steady annual immigration from large, unfragmented public forest tracts (> 70% forest cover, such as the Ozark Mountains and northern Wisconsin) that functioned as regional demographic sources (λ₁ > 1).
• Annual Plants in Desert Dune Systems (Cakile edentula, Keddy 1981–1982): Along coastal sand dunes in Nova Scotia, Paul A. Keddy demonstrated that the annual sea rocket (Cakile edentula) achieves high seed production on the seaward beach strandline (source, where B₁ ≫ D₁), whereas plants growing in the middle and landward dunes suffer high mortality and low fecundity (λ₂ < 1). Wind-driven seed dispersal from the seaward strandline continuously restocks the landward dune sink populations every autumn.
• Alpine Butterflies and Checkerspot Metapopulations (Boughton 1999): In rocky outcrop and meadow patches occupied by the Edith's checkerspot butterfly (Euphydryas editha), David A. Boughton experimentally demonstrated both source–sink transitions and pseudo-sink dynamics triggered by frost events that eliminated host plants in otherwise preferred meadows.

How can a habitat with the highest population density be a demographic sink?

As Beatrice Van Horne (1983) and H. Ronald Pulliam (1988) showed, dominant territorial adults in a high-quality source habitat often defend large territories, capping the breeding density K₁ in the source while generating a large annual surplus of juveniles (λ₁ − 1). When these surplus subadults are excluded from the source and congregate in non-territorial flocks or smaller territories within a lower-quality sink where mortality is only slightly higher than reproduction (1 − λ₂ is small), the equilibrium sink density n̂₂ = K₁(λ₁ − 1)/(1 − λ₂) can substantially exceed the source density.

What is the difference between a true sink and a pseudo-sink?

A true sink has a per-capita growth rate below replacement even at near-zero population density (λ₂(0) < 1) and will go extinct if immigration is cut off. A pseudo-sink (Watkinson & Sutherland 1995) has a positive intrinsic growth rate at low density (λ₂(0) > 1) and a positive local carrying capacity K₂, but heavy immigration from a nearby source inflates its population above K₂ so that density-dependent mortality exceeds local births (λ₂(n̂₂) < 1); if isolated from immigrants, a pseudo-sink does not go extinct but stabilizes at K₂.

Why do source–sink dynamics inhibit local evolutionary adaptation at the edge of a species' geographic range?

At the periphery of a species' geographic range, marginal habitats often act as demographic sinks (λ₂ < 1) sustained by recurrent immigration from core source populations (λ₁ > 1). Because the number of immigrants arriving each generation from the core is large relative to the number of locally produced offspring that survive in the sink, alleles adapted to the core source environment continually swamp out locally beneficial mutations in the peripheral sink, maintaining evolutionary niche conservatism.

1. Pulliam, H. R. (1988). "Sources, Sinks, and Population Regulation." The American Naturalist, 132(5), 652–661.
2. Van Horne, B. (1983). "Density as a Misleading Indicator of Habitat Quality." The Journal of Wildlife Management, 47(4), 893–901.
3. Watkinson, A. R., & Sutherland, W. J. (1995). "Sources, Sinks and Pseudo-Sinks." Journal of Animal Ecology, 64(1), 126–130.
4. Robinson, S. K., Thompson, F. R., Donovan, T. M., Whitehead, D. R., & Faaborg, J. (1995). "Regional Forest Fragmentation and the Nesting Success of Migratory Birds." Science, 267(5206), 1987–1990.
5. Dias, P. C. (1996). "Sources and Sinks in Population Biology10037-9)." Trends in Ecology & Evolution, 11(8), 326–330.