Source-Sink Dynamics

In population ecology, every local population's size changes according to the BIDE framework: ΔN = B − D + I − E where B = births, D = deaths, I = immigrants, E = emigrants, over some time interval. A source population is one in which local reproduction exceeds local mortality — B > D — so that, considered in isolation (no dispersal), the patch would grow: its intrinsic finite rate of increase, λ, exceeds 1. Because it produces more offspring than needed to replace itself, a source generates a surplus of dispersing individuals, making it a net exporter. A sink population, conversely, has D > B — its isolated λ is less than 1 — so the local population would decline toward extinction without a continual subsidy of immigrants. It is a net importer. Crucially, the distinction between source and sink is about the vital rates (birth and death rates, and hence λ), not about the standing population size N. A sink patch can, in principle, hold more individuals than a source patch at any snapshot in time, especially if it is large in area, superficially attractive, or simply positioned to receive heavy immigration. This decoupling is the conceptual crux of the theory and the reason it is taught as an advanced elaboration of metapopulation biology: naive Metapopulations models often assumed patches were demographically equivalent or that occupancy alone signaled quality, whereas source-sink theory insists on measuring the components of λ separately in each patch. The theory further distinguishes true (or "black-hole") sinks, which cannot sustain a positive population under any realistic immigration rate because deaths so outstrip births that immigration would have to be infinite, from ordinary sinks that persist stably at some finite equilibrium size fed by a steady inflow. It also connects to the ideas of habitat selection and the ideal free distribution: if animals cannot correctly assess patch quality (an "ecological trap"), they may settle in sinks at rates that make the trap self-reinforcing, a phenomenon of direct conservation concern.

Step 1 — Establish patch-specific vital rates. Each patch i has its own birth rate b_i and death rate d_i, set by local resources, predation, competition, or microclimate. In isolation, the local finite rate of increase is λ_i = 1 + b_i − d_i (in a simple discrete-time formulation). If λ_i > 1, patch i is a source; if λ_i < 1, it is a sink. These rates arise from ordinary demographic mechanisms — food abundance affecting fecundity, predator density affecting mortality, nest-site availability affecting recruitment — nothing exotic, only patch-to-patch heterogeneity in ordinary parameters.

Step 2 — Surplus production and dispersal. In a source, each generation more offspring are produced than the patch's carrying capacity can retain. The "extra" individuals — those who cannot secure a breeding spot, territory, or resource in the source — are forced out or actively disperse. This surplus is the mechanistic origin of emigration: it is not a mystical "decision" toward altruism but a direct, density-dependent consequence of local recruitment exceeding local carrying capacity, often mediated by aggressive exclusion of subordinate or young individuals by established residents (a mechanism documented across birds, rodents, and fish).

Step 3 — Immigration into sinks. Dispersing surplus individuals move across the landscape (constrained by dispersal distance, corridors, and habitat matrix permeability) and some fraction settles in sink patches. Settlement in a sink can happen for several non-exclusive reasons: (a) all source patches are already saturated at carrying capacity, so surplus individuals have no choice but to try elsewhere; (b) the sink is perceptually similar to a source (same vegetation structure, similar cues) so animals cannot distinguish quality — an ecological trap; (c) territorial exclusion by dominant conspecifics forces subordinates into worse habitat (despotic distribution, contrasted with the ideal free distribution in which individuals should settle in proportion to resource quality).

Step 4 — Local demographic failure masked by continual inflow. Inside the sink, births still occur but do not replace deaths; without immigration the local population would decay geometrically toward zero at rate λ_sink each generation. Continuous immigration refills the deficit, and if the immigration rate exceeds the demographic deficit (I ≥ D − B), the sink population can reach and hold a stable, sometimes large, equilibrium size — indistinguishable in a simple census from a healthy, self-sustaining population.

Step 5 — Regional (metapopulation-level) equilibrium. At the scale of the whole landscape, the metapopulation persists if total births across all source patches exceed total deaths across the whole system, i.e., the sum of source surplus must be large enough to cover all sink deficits plus dispersal mortality. This is the regional analogue of λ > 1, and it is what should be managed and monitored — not any single patch's population size.

The theoretical foundation was laid by H. Ronald Pulliam in a landmark 1988 paper, "Sources, Sinks, and Population Regulation" (The American Naturalist). Pulliam built a simple two-patch demographic model in which a source patch produced a fixed annual surplus of juveniles (because adult survival and territory number were fixed, but fecundity exceeded replacement) and a sink patch, where local reproduction failed to replace local mortality, absorbed the surplus. The mathematical result — now a textbook cornerstone — was that a sink population can attain a large, completely stable equilibrium size, dependent only on the rate of immigration from the source, even though its own vital rates would drive it to extinction in isolation. Pulliam also showed that under certain conditions the sum of sink populations across a landscape can exceed the population in the source, meaning a naive census-based ranking of "important habitat" could rank sinks above the very source that sustains the whole system — a direct warning against inferring habitat quality from abundance alone. The clearest empirical test came from work on migratory songbirds breeding in fragmented forest landscapes, particularly studies compiled and extended by Pulliam and colleagues (and, in a related independent line, by Donovan, Thompson, Faaborg, and Robinson in the early 1990s) on Neotropical migrant songbirds in Midwestern forest fragments. Researchers measured nest success and fledgling production directly — rather than simply counting adult birds — in forest fragments of varying size and edge exposure. They found:

• Large, contiguous forest tracts had high nest success (often >50%) and produced surplus young — behaving as sources.
• Small, fragmented woodlots, despite often holding dense breeding populations of the same species, suffered heavy nest predation and brood parasitism (notably by Brown-headed Cowbirds, Molothrus ater) along habitat edges, driving nest success below the level needed to replace adult mortality — behaving as sinks.
• Radio-tracking and banding data showed continual influx of young birds from source fragments into sink fragments, exactly as the Pulliam model predicted, sustaining apparently healthy breeding densities in patches that were reproductive failures.

A second influential empirical case is the Kirtland's Warbler (Setophaga kirtlandii), whose population was long known to depend almost entirely on a narrow band of young jack pine forest in Michigan maintained by fire or forestry — essentially a managed source — while occasional breeding attempts in other habitat types functioned as sinks subject to heavy cowbird parasitism, a pattern that directly informed a management program of cowbird control and habitat rotation. A complementary line of evidence comes from experimental and observational work on the ideal free distribution and habitat selection (building on Fretwell and Lucas's 1969–70 theoretical work, which Pulliam's model explicitly extends). Studies quantifying reproductive success against measured settlement density in species from great tits to stream fish have repeatedly shown cases where individuals settle in apparent proportion to habitat attractiveness cues rather than actual fitness payoff, producing ecological traps — a phenomenon later demonstrated experimentally, for instance in studies where reduced nest predation risk was manipulated to test whether birds could adjust settlement patterns (they often could not fully compensate), reinforcing that dispersal decisions do not perfectly track true patch quality. Together, these lines of evidence — Pulliam's formal model, direct measurement of source and sink demographic rates in songbirds, and studies of maladaptive habitat selection — established that (1) source-sink structure is real and detectable via BIDE accounting, (2) it is common in fragmented landscapes, and (3) census-based habitat assessment without demographic data is unreliable and can actively mislead conservation priorities.

Example 1 — Detecting a hidden sink from BIDE data. A wetland bird species is monitored in two patches over one breeding season.

Let's compute end-of-season N using ΔN = B − D + I − E. Patch A: ΔN = 140 − 60 + 10 − 90 = 0. N stays at 200. Isolated balance (ignoring I/E): B − D = 140 − 60 = +80 → strong surplus. Patch A is a source: it produced 80 more offspring than deaths, and exported 90 individuals (its surplus plus a bit more, since it also received 10 back), remaining stable at 200 purely through this export/import near-balance. Patch B: ΔN = 40 − 100 + 90 − 5 = 25. N rises from 150 to 175. Isolated balance: B − D = 40 − 100 = −60. Patch B is a sink: births fall 60 short of covering deaths. Yet because it received 90 immigrants (far more than needed to cover the 60 deficit) while losing only 5 emigrants, its census-visible population actually grew by 25 individuals this season. A field biologist who only recorded N at the start and end of the season would see Patch A "flat" and Patch B "growing" and might mistakenly rank Patch B as better habitat. Only decomposing the change into its B, D, I, E components reveals that Patch B is a demographic sink being kept afloat — and growing in visible size — by heavy immigration from Patch A.

Consequence: if Patch A (the actual source) were destroyed, Patch B would lose its 90 immigrants/season. Its local deficit of −60 would then be unmasked, and absent immigration N would decline geometrically. If we approximate λ_sink ≈ (150 + 40 − 100)/150 ≈ 0.60 per season for Patch B alone, then after loss of the source: N_t = 150 × (0.60)^t

Patch B collapses toward extinction within about 4–5 seasons, despite having looked like the "growing," healthier population the year before.

Example 2 — Regional persistence threshold. Suppose a metapopulation has one source patch producing a fixed annual surplus S = 120 dispersers, and two sink patches, each with an isolated deficit (D − B) of 45 individuals per year, and each losing an additional 10% of arriving immigrants to dispersal mortality (predation/exposure while moving between patches). Required gross immigration into each sink to exactly replace its deficit: Effective immigrants needed = deficit / (1 − mortality) = 45 / 0.90 = 50 dispersers arriving per sink per year. Total dispersers needed for both sinks = 2 × 50 = 100. The source supplies S = 120, so after feeding both sinks (using 100), a surplus of 20 dispersers per year remains — available to found new patches, buffer stochastic bad years, or be lost to dispersal mortality elsewhere in the landscape. The regional metapopulation is persistent because total source output exceeds total sink demand: 120 (supply) > 100 (demand) → regional λ_total > 1. If a third sink patch of identical deficit (45, needing 50 effective immigrants) were added — say through further habitat fragmentation — total demand becomes 150 dispersers/year, exceeding the source's supply of 120. Now the system as a whole is regionally in deficit; all three sinks (and eventually the source itself, if sinks draw down dispersers that would otherwise recruit locally) face decline. This illustrates the key management insight: the number and severity of sinks a single source can support is finite and calculable, and landscape fragmentation that adds sinks faster than it protects or restores source capacity pushes the whole regional population toward collapse — well before any single patch's census would show a warning sign.

Conservation reserve design. Identifying which patches in a protected-area network are true sources is now a standard part of recovery planning for endangered species; misallocating protection to sink habitat (because it holds visible individuals) is a recognized planning failure mode that agencies explicitly try to avoid using demographic monitoring rather than census-only surveys.

Agricultural pest and biocontrol management. Crop fields often act as sinks for natural enemies of pests (or vice versa for the pests themselves), and understanding source patches — field margins, hedgerows, adjacent wild habitat — guides where to invest in habitat enhancement to sustain beneficial insect populations that suppress pests across a farmed landscape.

Marine reserve and fisheries design. Marine protected areas are frequently designed explicitly as sources: no-take zones that build up spawning stock biomass whose larval or adult output ("spillover") sustains fished sink areas outside reserve boundaries, a direct real-world application of Pulliam-style dispersal subsidy.

Urban and roadside ecological traps. Streetlights, reflective surfaces, and mowed roadside verges can act as sinks or outright ecological traps for insects, turtles, and birds that are behaviorally attracted to cues that no longer predict safety or resources, informing wildlife-crossing and lighting-mitigation policy.

Epidemiology and reservoir dynamics. The source-sink logic generalizes directly to infectious disease ecology, where "reservoir" host populations (sources of pathogen) sustain spillover infections in "dead-end" host populations (sinks) that cannot sustain transmission alone — a framework central to zoonotic spillover research (e.g., bat-to-human pathogen studies).

Foreshadowing human demography. The same BIDE bookkeeping — births, deaths, immigration, emigration — that separates sources from sinks in wildlife populations is exactly the accounting framework used to analyze regional and global Human Population Growth, where migration between regions of differing fertility and mortality plays a structurally similar role in shaping where human population growth is concentrated.

Ask the tutor: How would you redesign a habitat survey protocol to distinguish a true source from a heavily-immigration-subsidized sink using only two field seasons of data?