How Do Ecologists Estimate Population Size?

Ecologists estimate animal population sizes using mark-recapture sampling techniques that track marked individuals over discrete capture events. This guide outlines the core mathematical models, including the Lincoln-Petersen index and open population estimators like the Jolly-Seber model, while detailing key biological assumptions and field constraints.

Fundamentals of Mark-Recapture Sampling

Mark-recapture sampling, sometimes called capture-mark-recapture (CMR), is a cornerstone methodology in field biology and wildlife management. In its simplest form, biologists capture a sample of individuals from a wild population, mark them with identifiable tags or bands, release them back into the wild to mix, and subsequently capture a second sample. By calculating the proportion of marked animals recovered in the second sampling event, researchers can mathematically infer total population abundance.

The Lincoln-Petersen Estimator for Closed Populations

The foundational framework for mark-recapture studies is the Lincoln-Petersen estimator, designed for closed populations where birth, death, immigration, and emigration do not occur during the study interval.

Mathematical Formulation

The estimator relies on a simple ratio: the proportion of marked animals in the second sample should equal the proportion of the total marked population within the entire community.

\[ \frac{M}{N} = \frac{R}{C} \]

Solving for population size (\(N\)):

\[ N = \frac{M \times C}{R} \]

Where:

  • \(M\) is the number of individuals marked and released during the first capture occasion.
  • \(C\) is the total number of individuals captured in the second sampling occasion (both marked and unmarked).
  • \(R\) is the number of marked individuals recaptured in the second sampling occasion.
  • \(N\) is the estimated total population abundance.

Chapman Modification

In field conditions with small sample sizes or low recapture rates, the basic Lincoln-Petersen estimator produces a biased estimate of \(N\). To account for small-sample bias, D.G. Chapman introduced a corrected formula that produces an unbiased estimate when \(M + C \ge N\):

\[ N = \frac{(M + 1)(C + 1)}{R + 1} - 1 \]

The associated variance can be estimated using:

\[ \text{Var}(N) = \frac{(M + 1)(C + 1)(M - R)(C - R)}{(R + 1)^2 (R + 2)} \]

Core Biological Assumptions

For closed population estimators to yield reliable abundance estimates, five critical assumptions must be satisfied:

  1. Population Closure: The population is physically and demographically closed between sampling periods; no individuals enter via birth or immigration, and none leave through death or emigration.
  2. Equal Catchability: Every individual, marked or unmarked, has an equal probability of capture during each sampling event.
  3. No Mark Loss: Marks, tags, or passive integrated transponders (PIT tags) remain intact and readable throughout the survey duration.
  4. Complete Mark Reporting: Field biologists detect and accurately record all recaptured marks without error.
  5. Normal Behavior and Survival: The tagging or handling process does not alter an individual's behavior, catchability (e.g., trap-happiness or trap-shyness), or probability of survival.

Multi-Sample Models for Closed Populations

When field protocols permit more than two sampling sessions, closed multi-capture estimators provide higher statistical precision and permit relaxation of the equal-catchability assumption.

Schnabel Method

The Schnabel method extends the Lincoln-Petersen approach to repeated sampling occasions over a short period. In this formulation, unmarked individuals captured on each trip are marked and released, steadily increasing the cumulative pool of marked individuals (\(M_t\)).

The population estimate is calculated using weighted linear regression approximations:

\[ N = \frac{\sum_{t=1}^{s} (C_t M_t)}{\sum_{t=1}^{s} R_t} \]

Where:

  • \(s\) is the total number of sampling occasions.
  • \(C_t\) is the total number of animals caught in sample \(t\).
  • \(R_t\) is the number of marked animals recaptured in sample \(t\).
  • \(M_t\) is the total number of marked animals present in the population just prior to sample \(t\).

Program MARK and Heterogeneity Models

Modern wildlife studies commonly deploy generalized linear models using software packages such as Program MARK. These handle complex variations in capture probability:

  • Model \(M_0\): Constant capture probability across all individuals and occasions.
  • Model \(M_t\): Capture probability varies across time or sampling occasions (e.g., changing weather conditions).
  • Model \(M_b\): Behavioral response to capture, accounting for trap-happy or trap-shy responses.
  • Model \(M_h\): Individual heterogeneity, where capture probability varies inherently among individuals based on sex, age, or territory placement.

Open Population Models

Field studies conducted across extended time horizons inevitably violate demographic and geographic closure assumptions. Under these conditions, ecologists transition to open population models.

The Jolly-Seber Model

The Jolly-Seber (JS) model evaluates open populations by simultaneously estimating population size (\(N_t\)), apparent survival probabilities (\(\Phi_t\)), and the addition of new individuals through recruitment (\(B_t\)).

Because the model explicitly calculates the survival rate from occasion \(t\) to \(t+1\), it decouples mortality and emigration from true population abundance fluctuations.

Cormack-Jolly-Seber (CJS) Model

When estimating abundance is secondary to evaluating demographic rates, biologists utilize the Cormack-Jolly-Seber (CJS) model. The CJS model conditions entirely on the first capture and focuses strictly on estimating survival (\(\Phi\)) and recapture probabilities (\(p\)), offering robust survival parameters without demanding strict whole-population enumerations.

Modern Advances: Spatially Explicit Capture-Recapture

Traditional mark-recapture techniques often assume that individuals are evenly distributed across the sampling landscape. Spatially Explicit Capture-Recapture (SECR) incorporates spatial location data from GPS telemetry, camera traps, or non-invasive genetic tagging (hair snags and scat sampling).

SECR models estimate population density (\(D\)) directly by modeling animal home ranges as spatial activity centers:

\[ p(d) = g_0 \exp\left(-\frac{d^2}{2\sigma^2}\right) \]

Where:

  • \(p(d)\) is the capture probability at distance \(d\) from the animal's activity center.
  • \(g_0\) is the baseline capture probability when a trap is located at the center of the home range.
  • \(\sigma\) defines the spatial scale parameter governing home-range movement.

By decoupling trap layout from capture probability, SECR eliminates boundary edge effects and provides biologically realistic density estimates across diverse ecosystems.