Greater Sage-Grouse Population Monitoring Framework: Glossary of Terms

Click on a word below to scroll directly to its definition.

Terms related to leks and population clusters

• Lek
• Cluster
• Neighborhood Cluster
• Climate Cluster
• Least-Cost Path Minimum Spanning Trees (LCP–MST)
• Clustering Algorithm
• Thiessen Polygons

Terms related to population trend estimation

• Apparent Abundance (\( \hat{N} \))
• State-Space Model (SSM)
• Abundance Index
• Finite Rate of Population Change (lambda, \( \hat{λ } \))
• Average Annual Rate of Change
• Intrinsic Rate of Population Change (\( \hat{r } \))
• Prior Distribution
• Posterior Distribution
• Credible Interval
• Oscillation
• Cycle
• Abundance Nadir
• Time Period
• Trend
• Extirpation Probability

Terms related to the Targeted Annual Warning System (TAWS)

• Aberrant Decline
• Signal (TAWS)
• Slow Signal (TAWS)
• Fast Signal (TAWS)
• Watch^r (TAWS)
• Warning^r (TAWS)
• Chronic Warning^N (TAWS)

Lek

Technical Definition: A traditional, site-faithful breeding arena characteristic of species that congregate in areas for mating opportunities. In greater sage-grouse (Centrocercus urophasianus), leks are open, relatively flat areas where males congregate each spring to perform elaborate strutting displays. Lek locations are persistent across years, which allows them to serve as fixed, repeatable sampling units for population monitoring.

Data provided from state wildlife agencies may include observations of signs and no birds, so only leks with ≥2 displaying males observed for ≥2 years are considered as leks. The maximum number of male sage-grouse observed during spring counts at leks provides the primary data input for the state-space model (SSM), which generates estimates of population trends, extirpation probabilities, and Targeted Annual Warning System (TAWS) alerts.

General Definition: A lek is a traditional site — used year after year at the same location — where male sage-grouse assemble each spring to perform their elaborate mating displays and compete for females. Because the same sites are revisited annually, leks provide a reliable, consistent location for counting birds and tracking population performance over time. The counts from these sites are the primary data underlying all population analyses.

Management Implications: Provides the core data used to track population change. Consistent annual survey effort at known leks is essential to the reliability of all downstream analyses, including trend estimation, extirpation probabilities, and TAWS alerts.

Back to top.

Cluster

Technical Definition: A spatially delineated, biologically based management unit formed by grouping leks according to sage-grouse habitat use, landscape connectivity, and ecological similarity. The hierarchical monitoring framework employs 13 nested spatial levels, ranging from smaller local groupings to the broadest regional groupings. Note that while individual leks represent the finest resolution of population structure, neighborhood clusters (level 2) constitute the finest cluster scale used in population modeling and trend estimation, and climate clusters (level 13) are the broadest.

This nested architecture allows population status to be assessed simultaneously at multiple spatial resolutions. Two cluster levels — neighborhood clusters and climate clusters — are specifically used for aggregating lek count data that informs estimates of population trend, extirpation probabilities, and TAWS alerts.

General Definition: A cluster is a group of leks organized together based on how sage-grouse use the land and move across it. Clusters are nested geographic zones — smaller local groups sit inside larger regional ones, analogous to the structuring of human populations in which neighborhoods fit within a city, which fits within a state.

This layered structure lets managers examine sage-grouse population performance at multiple scales at once, from a single local area all the way up to a broad region.

Management Implications: Allows consistent reporting and decision-making across spatial scales. Managers can evaluate population status at local, intermediate, and regional levels within a single, unified framework.

Back to top.

Neighborhood Cluster

Technical Definition: A mid-scale population cluster unit (level 2) in the hierarchical framework, defined by lek connectivity and abiotic/biotic similarity. Although individual leks represent the finest resolution of population structure, the neighborhood cluster is the finest spatial scale at which population clustering is applied in the monitoring model. Leks are assigned to the same NC when: (1) leks occur <15 km apart; (2) the connecting paths do not cross major roads with ≥4,000 annual average daily traffic (AADT); and (3) the abiotic and biotic conditions associated with leks show greater similarity to each other than to leks in other potential clusters.

NCs typically contain 20–30 leks, though smaller units occur when habitat conditions or connectivity rules prevent grouping. NCs are treated as relatively closed demographic units and serve as the primary local-scale units for trend estimation and TAWS risk assessment.

General Definition: A neighborhood cluster (NC) is a local group of leks — typically 20 to 30 leks — that are close to one another (within about 30 km), share similar habitat conditions, and are not separated from each other by major highways.

NCs represent the local scale at which sage-grouse population trends are tracked and TAWS assessments are conducted — the scale most directly relevant to on-the-ground management decisions.

Management Implications: Represents demographically meaningful, relatively closed populations for local-scale trend and risk assessment. Neighborhood clusters are one of the two primary spatial scales within TAWS and are the most management-relevant unit for identifying and responding to localized population declines.

Back to top.

Climate Cluster

Technical Definition: The broadest spatial unit in the hierarchical framework (level 13), defined by lek connectivity rules and shared climatic regimes in which conditions similarly influence population dynamics. Climate clusters encompass multiple neighborhood clusters.

Within the monitoring framework, climate clusters serve two primary roles: (1) generating high-level, range-wide trend summaries; and (2) functioning as the reference/control population unit in the TAWS framework, against which local (lek or NC) population dynamics are evaluated to determine whether declines are climatically driven or locally anomalous.

General Definition: A climate cluster (CC) is the largest geographic zone in the monitoring framework — a broad region where sage-grouse populations tend to respond similarly to the same prevailing weather patterns.

Climate clusters are used for two purposes: summarizing big-picture population trends across the range and serving as the regional reference population in the Targeted Annual Warning System. Because weather-driven population fluctuations are expected to affect all populations within a climate cluster similarly, the CC provides the regional backdrop against which local trends are compared.

Management Implications: Provides a stable regional baseline against which local (lek or NC) declines can be evaluated. Serves as the reference population in TAWS, enabling managers to separate weather-driven regional declines from locally driven declines that may be responsive to management.

Back to top.

Least-Cost Path Minimum Spanning Trees (LCP–MST)

Technical Definition: A graph-theoretic construct used to identify biologically realistic connectivity pathways among points — in this application, active leks. In graph theory, a spanning tree is an acyclic subgraph that connects all points without forming loops. The minimum spanning tree (MST) is the spanning tree with the minimum total edge weight.

Edge weights (“costs”) represent landscape resistance to sage-grouse movement, incorporating inter-lek distance and other movement-related factors. Least-cost paths between lek pairs are computed across a resistance surface (developed from multiple scales of elevation and sagebrush cover (habitat), and avoidance of rugged terrain, large water bodies and inundated salt flats, and tree canopy cover), and the resulting network is pruned to its MST to identify the most parsimonious set of lek connections. The LCP–MST forms the structural backbone of the lek connectivity network, which is used as input to the SKATER clustering algorithm.

General Definition: The LCP–MST is a map-based network that identifies the most efficient pathways connecting all known active leks across the landscape, following routes that account for how easily sage-grouse can move through the surrounding terrain.

Each connection between leks is assigned a “cost” based on factors such as distance and landscape features that impede movement. The resulting network links every lek to the system in the most efficient manner possible, without redundant loops — providing the spatial foundation for grouping leks into biologically coherent population clusters.

Management Implications: Helps define biologically realistic population boundaries by accounting for landscape features that facilitate or impede sage-grouse movement. Underpins the spatial structure of neighborhood and climate clusters.

Back to top.

Clustering Algorithm

Technical Definition: A hybrid approach that used landscape connectivity of active leks (Least-Cost Path Minimum Spanning Tree [LCP-MST]), constraint-based rules (literature on greater sage-grouse movement distances, loss of connectivity due to high-use transportation networks, and suggested number of leks to cluster for each hierarchically nested cluster/population level), and a spatial clustering algorithm to define populations for each subgraph of the LCP-MST. The LCP-MST is first decomposed into subgraphs according to constraint-based rules at each hierarchical level to inform the development of nested population clusters. Second, a clustering algorithm is applied to individual subgraphs of each hierarchical level. The Spatial “K”luster Analysis by Tree Edge Removal (SKATER) clustering algorithm partitions spatially referenced data into homogeneous groups by iteratively pruning edges from a minimum spanning tree (e.g., LCP-MST). At each step, the algorithm identifies the edge whose removal maximizes within-cluster homogeneity and maximizes between-cluster heterogeneity. Homogeneity/heterogeneity are based on the LCP-MST edge weights and habitat covariates of leks. This process repeats until the suggested range of leks, covariates, and selection of edge weights yields the best-performing model. The process is then repeated for each subgraph and for all combinations of covariates and scales (ranging from single cell to 6,400 meters, depending on the source data) until the model with the lowest AIC is identified (thousands of models evaluated per cluster level). In the sage-grouse application, the LCP–MST connects leks based on conditions that restrict movement, and the clustering algorithm is informed by covariates, including terrain, vegetation, and 30-year climate-averages to produce ecologically coherent, spatially contiguous lek groupings across the 13 hierarchical levels.

General Definition: Defining greater sage-grouse clusters/population units relied on connectivity (proximity) of leks, rules to ensure a spatially balanced number of leks within population clusters, and similar habitat and climate conditions among leks. The result is a set of population units that are both geographically contiguous and ecologically coherent. The methodology transforms a complex network of lek connections into an organized, nested cluster hierarchy used throughout the monitoring framework.

Management Implications: Ensures population units reflect both spatial proximity and biological similarity, producing ecologically meaningful groupings that support defensible, reproducible monitoring and management decisions.

Back to top.

Thiessen Polygons

Technical Definition: Thiessen polygons are a spatial partitioning method derived from a set of seed points in which each polygon contains all locations closer to its associated seed point than to any other. In this framework, seed points correspond to individual, active leks assigned to clusters. The tessellation forms a seamless mosaic with no gaps or overlaps, completely partitioning the spatial domain.

Thiessen polygons delineate geographically continuous catchment regions around lek clusters and facilitate assignment of newly discovered leks to existing cluster units by identifying which cluster’s seed leks are nearest. Polygon boundaries are equidistant between adjacent seed leks in different clusters and do not represent habitat suitability boundaries.

General Definition: Thiessen polygons are map zones drawn so that every point on the landscape belongs to the territory of the nearest lek cluster. Each zone extends outward from its associated leks to the midpoint between those leks and neighboring cluster leks, creating a continuous mosaic with no gaps.

Their primary practical use is placing newly discovered leks into the appropriate population group. These are spatial bookkeeping boundaries, not habitat maps — a polygon may include areas with no suitable sage-grouse habitat, such as urban areas or water bodies.

Management Implications: Ensures consistent spatial units for monitoring across the entire landscape, including areas of potential future range expansion. Provides a rule-based, reproducible method for incorporating newly detected leks into the monitoring framework.

Back to top.

Apparent Abundance (\( \hat{N} \))

Technical Definition: The model-estimated number of male sage-grouse attending leks within a defined population unit (individual lek or NC), derived from the state-space model (SSM). Estimates are characterized as “apparent” rather than true abundance because they are conditioned on the behavior of lek-attending males and do not account for females, non-attending males, or individuals at unmonitored sites.

This metric is conceptually analogous to a detection-limited index of relative abundance — a proxy for population size rather than a census estimate. Because apparent abundance is derived consistently from the same sampling protocol (male lek counts) and the same model structure across all spatial units and time periods, it provides a standardized basis for trend estimation, TAWS signal detection, and extirpation probability calculation. Relative change in apparent abundance captures the information needed for all three applications, and true population abundance is not required for these analyses.

General Definition: Apparent abundance is the model-based estimate of how many male sage-grouse are present at a lek or population cluster in any given year. It is called “apparent” because only males attending leks during surveys are counted — females and males that do not attend are not included.

Although it does not represent a complete count of all birds, apparent abundance is derived consistently across all sites and years, making it a reliable and comparable index for tracking population change over time.

Management Implications: Provides a consistent, model-based estimate of male numbers over time that can be compared across leks, clusters, and regions for trend and risk assessment. The consistency of this metric across time and space is essential for detecting meaningful population change.

Back to top.

State-Space Model (SSM)

Technical Definition: A class of hierarchical statistical models that decompose observed time-series data into a latent (unobserved) process component and an observation component. The state (or transition) equation describes the true but unobserved population dynamics — the actual trajectory of male apparent abundance over time — while the observation equation links the latent state to observed lek counts, explicitly accounting for observation error and missing survey data.

In this application, the SSM is implemented within a Bayesian hierarchical framework, enabling estimation of unobserved population states, uncertainty propagation via posterior distributions, and borrowing of statistical strength across related spatial units. Model-estimated apparent abundance values (\( \hat{N} \)) derived from the SSM underlie all downstream trend and TAWS analyses.

General Definition: The state-space model (SSM) is the statistical framework underlying all population estimates in this monitoring system. Lek surveys are imperfect. For example, some counts are conducted under poor weather conditions and some leks are not surveyed every year. The SSM uses all available count data together with mathematical rules about how populations change over time to produce reliable estimates of how many males were present each year.

Rather than reporting raw counts, the SSM provides the best available estimate of apparent abundance along with a quantified measure of uncertainty in that estimate. This approach allows for consistent comparisons across sites and years even when survey coverage is uneven.

Management Implications: Provides robust trend estimates even when field data are incomplete or variable. The model’s ability to separate biological signal from observation error makes it well-suited for long-term population monitoring across a vast landscape with variable survey coverage.

Back to top.

Abundance Index

Technical Definition: A dimensionless, relativized metric of population size derived from SSM-estimated apparent abundance (\( \hat{N} \)). Calculated as the ratio of each year’s estimated abundance to the long-term mean abundance across all years of the analysis (for example, 1960–2024). Because absolute \( \hat{N} \) values vary widely in magnitude across leks and clusters of different sizes, the abundance index standardizes all trajectories to a common scale (long-term mean = 1.0), enabling direct comparison of population dynamics across spatial units of differing absolute size.

Plain Language:

The abundance index puts populations of very different sizes on a common scale so they can be more easily compared. It rescales each population’s abundance relative to its own long-term average: a value of 1.0 means at the historical average; 0.5 means at half the average; 1.5 means 50% above average.

This allows any given year to be placed in context relative to the full monitoring history and allows a small local lek to be compared to a large regional cluster on equal footing.

Management Implications: Allows comparison of populations of different sizes on a common scale. Serves as the primary input to lambda (λ) calculations and trend analyses.

Back to top.

Finite Rate of Population Change (lambda, \( \hat{λ } \))

Technical Definition: A fundamental demographic parameter representing the multiplicative rate of change in population size between two consecutive annual time steps. Formally defined as \( \hat{λ } \)_t(t) = \( \hat{N} \)_t / \( \hat{N} \)_t-1 , where \( \hat{N} \) is the SSM-estimated apparent abundance. Values of \( \hat{λ } \) > 1 indicate population growth; \( \hat{λ } \) < 1 indicates decline; \( \hat{λ } \) = 1 indicates stable abundance.

Lambda is the discrete-time analog of the continuous intrinsic rate of increase (\( \hat{r } \)), related by \( \hat{λ } \) = e^r.

General Definition: Lambda (λ) is the year-over-year ratio of population size — it indicates whether a population grew, shrank, or held steady from one year to the next. A lambda of 1.2 means the population is 20% larger than the prior year; a lambda of 0.8 means it is 20% smaller; 1.0 means no change.

Values above 1.0 indicate growth; values below 1.0 indicate decline.

Management Implications: Provides an intuitive, annually updated measure of population change.

Back to top.

Average Annual Rate of Change

Technical Definition: The geometric mean annual rate of population change calculated between two successive abundance nadirs (across a defined time period). By anchoring calculations to nadirs rather than arbitrary calendar endpoints, this metric removes the confounding influence of intra-cycle oscillations and provides a cycle-adjusted measure of the underlying population trajectory.

Expressed as the mean \( \hat{λ } \) per year over the nadir-to-nadir interval, this metric can be compared across time periods and spatial units to characterize long-term directional trends independent of cyclical variation.

General Definition: The average annual rate of change is a smoothed, long-term estimate of how much a population typically grows or shrinks each year. By calculating it between population low points (nadirs), it removes the natural multi-year fluctuations that all sage-grouse populations exhibit.

The result is a trend signal that reflects the underlying long-term trajectory, not skewed by the phase of the population cycle at which the measurement begins or ends.

Management Implications: Provides a consistent, comparable measure of long-term population performance across leks, neighborhood clusters, and regions. Particularly useful for assessing whether populations have improved or declined across entire monitoring periods.

Back to top.

Intrinsic Rate of Population Change, \( \hat{r } \)

Technical Definition: The continuous-time per-capita rate of population change, defined as r = ln(λ), where λ is the finite rate of population change. Values \( \hat{r } \) > 0 indicate growth; \( \hat{r } \) < 0 indicate decline; \( \hat{r } \) = 0 indicates stability. In this framework, \( \hat{r } \) serves as the key parameter in the SSM’s state transition equation governing latent population dynamics.

In classical ecology, the intrinsic rate of natural increase (r) represents the per-capita population growth rate under ideal conditions: unlimited resources and absence of density dependence. As applied here, \( \hat{r } \) reflects the observed rate of change as captured through lek count data, incorporating the real-world constraints acting on the population rather than its theoretical maximum.

General Definition: The intrinsic rate of change (\( \hat{r } \)) expresses population growth or decline on a continuous mathematical scale rather than as a year-to-year ratio. Positive \( \hat{r } \) indicates growth; negative r indicates decline; zero indicates stability.

While lambda (\( \hat{λ } \)) describes the factor by which a population multiplied from one year to the next, (\( \hat{r } \)) expresses the same information as a continuously accruing rate, The two measures are mathematically equivalent and are related by the formula (\( \hat{r } \)) = ln(\( \hat{λ } \)).

Management Implications: Represents the fundamental rate of realized population change as estimated from monitoring data. Its log-scale properties make it well-suited for statistical modeling and for comparing rates of change across populations and time periods.

Back to top.

Prior Distribution

Technical Definition: In Bayesian statistical inference, the prior distribution encodes existing knowledge or uncertainty about an unknown model parameter before observed data are incorporated. Mathematically, the prior, P(θ), represents a probability distribution over possible values of the parameter θ — expressing the analyst’s state of knowledge before data are observed. Priors can be informative (derived from published studies, expert knowledge, or biological constraints) or weakly informative/diffuse (placing minimal constraints on parameter values, thereby allowing the data to dominate inference).

In the sage-grouse SSM, priors are specified for key parameters — such as the intrinsic rate of population change (\( \hat{r } \)) and process variance — drawing on estimates from prior monitoring analyses. Well-chosen priors improve model stability and ensure estimates remain biologically plausible, particularly for population units with sparse survey data.

General Definition: In Bayesian statistics, a prior distribution is a starting assumption built into the model before it processes any new survey data. It reflects existing knowledge — or the degree of uncertainty — about a value, such as how fast sage-grouse populations typically grow or decline based on decades of prior monitoring.

Once the model processes actual lek counts, it updates those starting assumptions. The prior serves as a reasonable anchor that keeps model estimates biologically plausible, particularly for leks or clusters with limited survey history.

Management Implications: Allows existing biological knowledge to inform and stabilize model estimates, particularly for population units with sparse data. Ensures that parameter estimates remain within biologically meaningful bounds even when survey data are limited.

Back to top. 

Posterior Distribution

Technical Definition: In Bayesian inference, the posterior distribution is the probability distribution of an unknown parameter conditioned on both the prior distribution and the observed data, derived via Bayes’ theorem: P(θ | data) ∝ ^{P(θ | data) x P(θ )}/ _P(data). This expression states that the posterior probability of θ given the data is proportional to the likelihood of the data given θ multiplied by the prior (unconditioned) probability of θ divided by the prior probability of the data.

For the sage-grouse SSM, posterior distributions of model parameters — including annual apparent abundance (\( \hat{N } \)) and trend (\( \hat{λ } \)) — are estimated using Markov chain Monte Carlo (MCMC) sampling. Some outputs, such as the abundance index and lambda, are derived parameters: quantities calculated directly from primary model parameters (\( \hat{N } \), \( \hat{r } \) at consecutive time steps) rather than estimated independently. All reported point estimates (e.g., medians) and uncertainty bounds (credible intervals) are summaries of these posterior distributions.

General Definition: After the model combines its starting assumptions (the prior) with actual field count data, the result is the posterior distribution — the model’s updated, data-informed estimate of what a parameter (such as population size or rate of change) most likely is.

The posterior is not just a single number, but a full range of plausible values and how probable each one is. The population estimates and credible intervals reported in this framework are summaries of these posterior distributions.

Management Implications: Provides the most informed estimate of population trends and the full range of uncertainty in that estimate. All reported trend values, abundance estimates, and extirpation probabilities in this framework are derived from posterior distributions.

Back to top.

Credible Interval (CRI)

Technical Definition: A Bayesian measure of parameter uncertainty. A credible interval (typically 95%) contains x% of the posterior probability mass, providing a direct probabilistic statement: there is an x% posterior probability that the true parameter value falls within the interval, given the model and observed data.

This interpretation is more direct than a Frequentist confidence interval, which conveys properties of a hypothetical repeated-sampling procedure rather than a probability statement about a specific parameter value. In this framework, CRIs quantify uncertainty around annual abundance estimates, trend estimates, and extirpation probabilities.

General Definition: A credible interval (CRI) is a range of values derived from the model’s output within which the true parameter value falls with a specified posterior probability — typically 95%. It provides a direct statement of uncertainty: there is a 95% posterior probability that the true value lies within the reported range, given the model and the data.

A narrow CRI indicates a precise estimate; a wide CRI indicates greater uncertainty, often because data are sparse for that lek or cluster. While conceptually similar to a confidence interval in traditional statistics, the CRI has a more direct probabilistic interpretation.

Management Implications: Communicates uncertainty around trend and abundance estimates. Helps managers understand not just what the trend estimate is, but how much confidence is warranted — a critical consideration when using model outputs to inform management decisions.

Back to top.

Oscillation

Technical Definition: In mathematics and physics, an oscillation is a periodic, repetitive variation of a quantity about a central value or between two states. In population ecology, oscillatory dynamics refer to recurring fluctuations in population size that follow a roughly regular temporal pattern rather than exhibiting monotonic growth or decline.

Greater sage-grouse populations exhibit well-documented oscillatory dynamics, with apparent abundance fluctuating at approximately 9-year intervals. These oscillations are driven primarily by climatic variables — particularly precipitation-mediated changes in vegetation structure and productivity — that drive multi-year variation in vital rates (survival and reproduction). The monitoring framework explicitly accounts for oscillatory dynamics by anchoring trend and extirpation analyses to population nadirs — prevents intra-cycle variation from biasing estimates of long-term directional change — and restricting spatially-nested comparisons of population change (\( \hat{r } \)) to the same year.

General Definition: Sage-grouse populations naturally rise and fall in a roughly 9-year rhythm, driven largely by wet and dry weather patterns that affect the sagebrush and other vegetation the species depends on for food, nesting, and raising young. This regular pattern of fluctuation is called an oscillation.

Recognizing this natural rhythm is important for interpreting population data. A declining count in any given year may simply reflect the normal downswing of the cycle rather than a locally driven change. The monitoring framework is specifically designed to distinguish these predictable natural fluctuations from declines that are decoupled from the regional pattern and may warrant management attention.

Management Implications: Helps distinguish natural population changes (up or down) from locally driven increases or declines. Trend analyses in this framework are anchored at abundance nadirs specifically to account for oscillatory dynamics, preventing natural cycle phases from being misinterpreted as long-term directional change. TAWS analyses restrict nested-population comparisons to a single, same year, nullifying confounding effects of oscillatory population dynamics.

Back to top.

Cycle

Technical Definition: In the context of sage-grouse population dynamics, a cycle is one complete oscillation in population abundance, measured from one abundance nadir to the next consecutive nadir. Because sage-grouse populations exhibit approximately 9-year oscillatory dynamics, a single cycle spans roughly 9 years.

General Definition: A cycle is one complete swing in the sage-grouse population — from one low point to the next. Because sage-grouse numbers naturally rise and fall on a roughly 9-year rhythm, a single cycle represents approximately one decade of population change.

Defining cycles this way provides a consistent, biologically grounded time unit for measuring long-term population trends.

Management Implications: Provides a biologically meaningful and statistically consistent time window for trend estimation. The nadir-to-nadir structure ensures trend comparisons are made at equivalent points in the population cycle.

Back to top.

Abundance Nadir

Technical Definition: The temporally identified minimum value of apparent abundance within a single population cycle, used as the reference anchor for nadir-to-nadir trend calculations. Nadirs are identified from the SSM-estimated abundance time series for each spatial unit (lek, NC, or CC).

General Definition: The abundance nadir is the lowest point in a population’s natural cycle — the year when estimated apparent abundance reaches its minimum before beginning to increase again.

Nadirs serve as the start and end points for calculating long-term trends, ensuring that trend estimates are not influenced by where the population happened to fall within its natural cycle at an arbitrary calendar date.

Management Implications: Serves as the anchor point for cycle-based trend calculations. Anchoring to nadirs ensures that trend estimates reflect genuine long-term directional change rather than artifacts of intra-cycle variation.

Back to top.

Time Period

Technical Definition: A structured, nadir-anchored temporal window composed of one or more population cycles used to summarize long-term population change. Time periods are indexed hierarchically for our greater sage-grouse framework: Period 1 spans the longest available sequence of consecutive nadir-to-nadir cycles; Period 2 spans the second-longest; and successively higher-numbered periods represent progressively shorter, more recent intervals.

As new nadirs are identified in ongoing monitoring data, a new, higher-numbered period is added, and each existing period gains one additional cycle — allowing the framework to extend forward in time without renumbering earlier periods. This architecture ensures trend estimates from different reporting years remain directly comparable.

Beginning with the 2025 reporting cycle, time periods are identified using a nadir-indexed naming convention in which each period is designated by the nadirs that define its start- and endpoints. For example, Period 1–2 spans the interval between the first and second identified nadirs (e.g., 1966–1975); Period 1–3 spans nadirs 1 through 3 (e.g., 1966–1986); and Period 3–6 spans nadirs 3 through 6 (e.g., 1986–2021). This system allows all possible nadir-to-nadir intervals to be explicitly represented, facilitating comprehensive trend comparisons across the monitoring history and a greater delineation of modern and historic population performance.

General Definition: A time period is a defined window of time — measured from population low point to population low point — used to calculate and report trends. The greater sage-grouse framework uses multiple overlapping windows simultaneously: from the full monitoring record (back to the 1960s) down to shorter, more recent intervals.

This gives managers both a long-range historical view and a near-term picture of how populations are performing. As monitoring continues and new low points are identified, new periods are added automatically, keeping the framework current without requiring past results to be recalculated.

Starting in 2025, periods are labeled using the nadir numbers that define their start and end points (e.g., Period 1–2 [e.g., 1966–1975] or Period 3–6 [e.g., 1986–2021]), making it straightforward to identify the timeframe each period represents.

Management Implications: Provides consistent, biologically meaningful windows for comparing long- and short-term trends across leks, neighborhood clusters, and regions. Multiple time periods allow managers to assess whether trends differ depending on the timeframe considered.

Back to top.

Trend

Technical Definition: The estimated directional change in population apparent abundance over a defined nadir-to-nadir time period, quantified as the average annual rate of change in the abundance index. Trend estimates are derived from SSM output for individual leks, NCs, and CCs across all defined time periods.

Measurement from nadir to nadir avoids over- or underestimating population performance during intra-cycle changes in population size that operate irrespective of longer-term trends. As new nadirs are identified in future monitoring data, additional trend periods are added and the trend record extends progressively, incorporating new lek count data from ongoing surveys.

General Definition: Trend describes the estimated direction and rate of change in a population’s abundance over a defined time period — whether the population has been increasing, decreasing, or remaining stable, and by how much. Trends are calculated between population low points (nadirs) to avoid being influenced by the natural ups and downs of the population cycle.

Trends are reported for multiple time windows — from short-term recent intervals to long-term historical ones — so managers can assess both the overall trajectory and what has occurred most recently. All estimates include credible intervals that convey the level of confidence in the direction and magnitude of change.

Management Implications: Provides long- and short-term views of population direction and magnitude. Trend estimates across multiple time periods allow managers to assess short-, medium-, and long-term population trajectories, supporting both near-term management responses and longer-term conservation planning.

Back to top.

Extirpation Probability

Technical Definition: The model-estimated probability that the apparent abundance of a lek or NC will decline to fewer than 2 males within a specified future time window, derived by projecting the posterior distribution of the SSM forward in time. The threshold of fewer than 2 males was selected as an operational criterion for functional extirpation — the point at which a lek is no longer viable as an active breeding unit — consistent with minimum-count thresholds used in state wildlife agency lek classification protocols.

The complement (1 − extirpation probability) yields the probability of persistence: the probability that the unit retains ≥2 males within the specified time window. Either metric can be used to characterize near- or long-term population viability.

General Definition: Extirpation probability is the estimated chance that a lek or population cluster will effectively cease to function as a breeding site — dropping to fewer than 2 males — within a specified future time window. The complement of this value, the probability of persistence, represents the estimated chance the site remains occupied.

Together, these metrics provide a forward-looking assessment of which populations face the greatest risk of functional loss, supporting prioritization of conservation resources.

Management Implications: Identifies populations at highest risk of functional loss and informs conservation prioritization. Provides a probabilistic, forward-looking basis for directing management resources toward populations most likely to be lost without intervention.

Back to top.

Aberrant Decline

Technical Definition: Within TAWS, an aberrant decline is a negative population rate of change at the lek or neighborhood cluster (NC) scale that is decoupled from the contemporaneous climate cluster (CC) rate of population change (CC can be increasing, stable or decreasing). Decoupling must meet TAWS’ divergence thresholds (categorized as Slow or Fast signals) and persistence criteria (e.g., Watch^r when Slow signals occur in 2 consecutive years; Warning^r when Slow signals occur in 3 of 4 years or Fast signals in 2 of 3 years). This construct distinguishes locally driven declines from broad, climate‑driven oscillations by benchmarking local change against the CC baseline and evaluating it annually.

General Definition: An aberrant decline is a drop in a local sage-grouse population (at a lek or neighborhood cluster) that is unusually steep as compared to what’s happening across the broader region (climate cluster). More simply, the local population is falling out of step with the normal, climate‑driven ups and downs seen at the regional scale.

Management Implications: An aberrant decline triggers TAWS categories (Watch^r, Warning^r, Chronic Warning^N) and prioritizes investigation of local stressors (e.g., habitat loss/fragmentation, disturbance, disease, or other site‑level pressures) that may be mitigated. It guides resource allocation for targeted monitoring and intervention, while separating climate‑driven, regional cycles (less actionable) from locally actionable declines, improving timeliness and effectiveness of conservation decisions.

Back to top.

Signal (TAWS)

Technical Definition: A TAWS-derived indicator generated on an annual basis when a local population (lek or NC) simultaneously meets two conditions: (1) the local population is declining (\( \hat{r} \) < 0); and (2) the local intrinsic rate of population change (\( \hat{r} \)) is decoupled from the concurrent CC \( \hat{r} \) — that is, the local rate of decline is greater in magnitude than the concurrent regional rate of change.

A signal is a single-year assessment; it does not require persistence across multiple years. Repeated signals across consecutive years serve as the basis for escalation to Watch and Warning status. Signals are classified as Slow or Fast based on the magnitude of decoupling between local and regional \( \hat{r} \) values.

General Definition: A signal is the most immediate TAWS indicator, generated in any year when a local population is declining more than the surrounding region would predict. It is a single-year assessment: a lek or neighborhood cluster either receives a signal in a given year, or it does not, based on that year’s rate of change relative to the regional trend.

Signals come in two categories — slow (modest divergence) and fast (pronounced divergence) — and a pattern of repeated signals across consecutive years forms the basis for escalation to Watch and Warning designations.

Management Implications: Serves as the most immediate indication of a locally anomalous decline in any given year. Repeated signals across consecutive years form the basis for escalating TAWS alert status.

Back to top.

Slow Signal (TAWS)

Technical Definition: A TAWS signal indicating that a local population (lek or NC) is declining and the magnitude of decoupling between the local intrinsic rate of population change (\( \hat{r} \)) and the concurrent regional (CC) \( \hat{r} \) is detectable but modest. Slow signals represent an early-stage anomaly consistent with an emerging, locally driven decline.

Two consecutive years of slow signals at the same lek or NC result in a Watch^r designation.

General Definition: A slow signal is an early-stage indicator: it means a local population is declining somewhat faster than the regional rate of population change, but the divergence is not yet pronounced. The monitoring system notes the pattern, but it does not by itself constitute a formal alert.

If a slow signal is observed in two consecutive years at the same location, it results in a Watch^r designation.

Management Implications: Early indication of a potential locally driven decline. Two consecutive slow signals at the same lek or NC result in a Watch designation.

Back to top.

Fast Signal (TAWS)

Technical Definition: A TAWS signal indicating that a local population (lek or NC) is declining and the magnitude of decoupling between the local intrinsic rate of population change \( \hat{r} \) and the concurrent regional (CC) \( \hat{r} \) is pronounced. Fast signals indicate a more severe or accelerating locally driven decline and carry greater urgency than slow signals.

General Definition: A fast signal is a stronger indicator than a slow signal: it means a local population is declining substantially faster than the surrounding region. The divergence from the regional rate of population change is pronounced enough to indicate a more serious local departure from expected conditions.

Fast signals move through the alert hierarchy more rapidly than slow signals because of the greater magnitude of divergence.

Management Implications: Indicates a more severe or rapidly developing locally driven decline. Fast signals escalate to Warning status more rapidly than slow signals.

Back to top.

Watch^r (TAWS)

Technical Definition: A TAWS alert level assigned to a population unit (lek or NC) when slow signals are detected in 2 consecutive years. A Watch represents the first formal, multi-year evidence of a sustained decoupled decline at the local scale and calls for heightened monitoring attention.

General Definition: A Watch is issued when a population has shown slow signals for two consecutive years. It indicates that the departure from the regional trend is becoming a consistent pattern rather than a single-year anomaly.

A Watch provides guidance for increased monitoring attention to the affected lek or cluster and may prompt investigation into local conditions that could explain the divergence from the regional trend.

Management Implications: Provides guidance for areas to closely monitor and evaluate for potential local stressors. A Watch^r designation prompts managers to review local conditions and assess potential factors that may be driving divergence from the regional trend.

Back to top.

Warning^r (TAWS)

Technical Definition: A TAWS alert level assigned to a population unit (lek or NC) when slow signals are detected in 3 of 4 consecutive years, or fast signals in 2 of 3 consecutive years. A Warning^r represents sustained, statistically aberrant decline relative to the regional trend — a pattern inconsistent with climate-driven variation alone.

A Warning^r may concurrently activate a chronic Warning^N. The superscript (r) designates that this alert is based on rate-of-change (\( \hat{r} \)) divergence from the regional trend, distinguishing it from the abundance-referenced chronic Warning^N designation.

General Definition: A Warning^r is issued when a population shows a persistent pattern of decline that is out of step with the regional trend: slow signals in 3 of 4 consecutive years, or fast signals in 2 of 3 consecutive years. This pattern indicates that something beyond normal weather-driven variation may be influencing the population.

A Warning^r designation indicates that the affected lek or cluster could benefit from evaluation of on-the-ground conditions and assessment of whether local factors may be contributing to the observed decline.

Management Implications: Indicates a sustained, locally anomalous decline. Provides guidance for areas that could benefit from management evaluation, including assessment of local habitat conditions and potential contributing factors.

Back to top.

Chronic Warning^N (TAWS)

Technical Definition: A TAWS alert category that activates concurrently alongside a Warning^r. Unlike the Warning^r — which is a binary, annual assessment based on rate-of-change (\( \hat{r} \)) divergence from the regional trend — the chronic Warning^N remains in effect beyond the year of Warning^r activation, providing an ongoing record of whether the population has demonstrated a meaningful rebound in abundance.

The chronic Warning^N is removed only once the population unit’s estimated abundance rises to match or exceed a projected recovery threshold. This threshold is derived from the climate cluster’s growth rate as applied to the population unit abundance, prior to the initial signal that lead to the Warning^N. This design ensures that a Warning is not prematurely lifted from population units that have ceased to decline (and decouple) but have not yet recovered to a biologically meaningful level of abundance that accounts for broader trends in population change.

General Definition: A chronic Warning^N is a designation that becomes active alongside a Warning^r and remains in effect until the population has demonstrated recovery to a meaningful level — specifically, until estimated abundance meets or exceeds a projected recovery threshold based on the regional trend.

While a Warning^r is evaluated each year based on the most recent rate of change, the chronic chronic Warning^N persists across years. It is removed only when the population’s abundance has recovered sufficiently relative to the regional trajectory — not simply when the rate of decline has slowed or stopped.

Management Implications: Prevents premature removal of warning status from population units that have not yet demonstrated a meaningful recovery in abundance. Ensures that management attention continues until recovery — not merely stabilization — is evidenced.

Back to top.

Co-production

We continue working with all collaborators to improve science, that is the foundation of these products and tools, which can help to inform sage-grouse management. Each year, we develop an updated version of a range-wide standardized lek count database to include new counts and historical corrections, and to improve data quality through rigorous quality-control and assurance methods. During this process, software is updated, peer-reviewed, and released to the public. The trends and TAWS models are then run annually with the most recent lek data provided by state partners. The results are released in a data series report and are then integrated back into the SagePop Tool for use by federal and state agency partners. We continue to inform partners of model updates and improvements and to facilitate the incorporation of new features and outputs into the application.

Data restrictions

State wildlife agencies collect and manage lek databases. Because sage-grouse are a species of conservation concern and sensitive to activities during breeding, these data are accessible only through state wildlife agencies.

Funders

U.S. Geological Survey (Ecosystem Mission Area, Land Management Research Program and Species Management Research Program; Wyoming Landscape Conservation Initiative) and the Bureau of Land Management.

Partners