In Biosecurity Commons, a risk map is a spatial layer that summarises the likelihood that a threat will become established across a region. At each grid cell, the model combines information about:

• Suitability of the non-living (abiotic) environment (for example, climate).
• Suitability of the living (biotic) environment (for example, habitat or hosts).
• The likelihood that a viable threat arrives at that location (propagule pressure).

The resulting map helps you visualise where conditions are most favourable for establishment, given what is known about the threat and its pathways.

Establishment likelihood (or establishment potential) is the probability that an exotic threat will become established at a particular location. It is governed by three fundamental spatial constraints:

• Abiotic suitability – the suitability of the non-living environment for the threat. This typically includes climate (too hot, too cold, too dry), terrain, and disturbance regimes. Key consideration: marginally suitable areas should not be ruled out too quickly, because species can adapt or be more tolerant than currently assumed.
• Biotic suitability – the suitability of the living environment for the threat. This includes host availability, habitat, food resources and other species interactions that could facilitate or hinder establishment. Key consideration: data on host distributions or host–threat relationships may be incomplete, so uncertainty should be acknowledged.
• Propagule pressure – also known as introduction or contamination rate. This describes how likely it is that a viable threat will arrive at a location, given all relevant pathways of entry and movement (for example, trade, human travel, transport networks, wind).

In Biosecurity Commons, establishment likelihood at each grid cell is approximated by multiplying these three components:

Here we give a more explicit description of how establishment likelihood is calculated in Biosecurity Commons. The goal is not to provide a fully formal statistical specification, but to show clearly how the main components fit together.

Events and probabilities at a single location

Consider a single grid cell (location) x. We define three events:

• A_x: abiotic conditions at x are suitable for the threat.
• B_x: biotic conditions at x are suitable for the threat.
• P_x: at least one viable propagule of the threat arrives at x.

Establishment at location x occurs if, and only if, all three events occur: the environment must be physically suitable, biologically suitable, and the threat must arrive. Let E_x be the event that the threat establishes at x. Then:

The true probability of establishment at x is therefore:

In principle, this could include complex dependencies between abiotic conditions, biotic conditions and propagule pressure. However, we rarely have enough data to fully specify these dependencies in biosecurity problems.

Working approximation used in Biosecurity Commons

To make the problem tractable, Biosecurity Commons uses an approximation that treats the three components as conditionally independent, given the location:

Each of these probabilities is not observed directly. Instead, they are estimated or constructed from models and data layers:

• Pr(A_x) – derived from a species distribution model or other climate suitability model that maps environmental covariates (for example, temperature, precipitation) to a probability or index of abiotic suitability.
• Pr(B_x) – derived from host / habitat data, land use or other biotic information, possibly translated into an index of “how suitable” the living environment is for the threat.
• Pr(P_x) – derived from pathway models, trade and movement data, interception records or expert judgement about the relative frequency of arrivals.

In many practical applications, these components are first created as scaled indices between 0 and 1, rather than strict statistical probabilities. For example, you might:

• Fit a species distribution model and rescale its output to lie between 0 and 1.
• Take a host density layer and rescale it so that the maximum value is 1 (most suitable) and 0 is unsuitable.
• Convert pathway volumes into a relative arrival intensity and rescale it between 0 (no arrivals) and 1 (highest observed arrival intensity).

These rescaled layers are then interpreted as probability-like indices and multiplied together to approximate relative establishment likelihood.

Weighting components

In some cases, you may want to give more weight to one component than another. For example, you might believe that propagule pressure is the dominant constraint for a particular threat (for example, a very climate-tolerant stored-product pest). In that case, a weighted version of the product can be used:

where α, β and γ are non-negative weights that reflect the relative importance of each component. When α = β = γ = 1, this reduces to the simple unweighted product. When, for example, γ > 1 and α, β < 1, propagule pressure plays a more dominant role in shaping the final pattern.

Scaling and normalisation

Because the product of three numbers between 0 and 1 is often very small, it is common to:

• Rescale the raw product so that it lies between 0 and 1 across the region. For example, divide by the maximum value across all cells.
• Optionally convert the continuous values into risk classes (for example, “very low”, “low”, “moderate”, “high”) using quantiles or expert-defined breakpoints.

This does not change the relative ordering of cells, but makes the resulting map easier to interpret.

A simple numeric example

Suppose that for a particular grid cell, the model outputs:

• Pr(A_x) = 0.8 (abiotic conditions are very suitable).
• Pr(B_x) = 0.5 (moderate host / habitat suitability).
• Pr(P_x) = 0.2 (relatively low arrival rate of the threat).

The unweighted (α = β = γ = 1) establishment likelihood is:

Here, even though the abiotic conditions are very suitable, the overall establishment likelihood is limited by relatively low propagule pressure and moderate biotic suitability. If the same threat had much higher propagule pressure in another region (for example, Pr(P_x) = 0.6), then:

which is three times higher, illustrating how differences in pathways can strongly affect relative risk.

From single locations to a risk map

The calculations above are done for every grid cell in the study area. The outputs are then assembled into a raster (grid) to form a risk map. On Biosecurity Commons, this is automated within the risk mapping workflow so that you only need to provide or select:

• Abiotic suitability layers or models.
• Biotic suitability layers.
• Propagule pressure layers or approximations.

The platform then combines these components according to the multiplicative framework described above, producing a reproducible establishment likelihood map that can be used in subsequent workflows.

A standard risk mapping workflow on Biosecurity Commons allows users to:

• Estimate abiotic suitability – typically via species distribution models built from occurrence data, climate layers and other environmental variables.
• Estimate biotic suitability – often using land use, vegetation and habitat datasets that reflect host and habitat availability.
• Estimate propagule pressure – using a mix of expert judgement, border surveillance and proxy data describing how threats move after arrival (for example, trade flows, passenger arrivals, road networks, ports).
• Combine these three likelihoods into an establishment likelihood map that is standardised, documented and reproducible.

On Biosecurity Commons, each of these steps is implemented as a transparent, parameterised stage in the workflow. This makes it easy to adjust assumptions, rerun analyses and share results with collaborators.

Risk maps play several roles in biosecurity decision-making by turning complex data into actionable information. For example, they can be used to:

• Prioritise surveillance efforts: focus monitoring in areas with the highest likelihood of establishment, improving early detection.
• Assess surveillance coverage: check whether current surveillance activities adequately cover high-risk locations.
• Estimate area freedom: support trade and market access by quantifying the probability that a region is free of a threat, given surveillance data, detection probabilities and establishment likelihood.
• Inform spread modelling: identify plausible initial establishment sites and constrain spread simulations using realistic abiotic and biotic suitability patterns.
• Support impact analysis: combine establishment likelihood with impact layers to estimate potential economic, environmental or social harm if a threat establishes.

Because risk maps on Biosecurity Commons are standardised and shareable, they help build a common evidence base that can be reused across workflows such as dispersal modelling, surveillance design, and proof of freedom analyses.

• A step-by-step guide to running a risk mapping experiment is provided in the  Risk Mapping Quick Start Guide .
• A short video overview of the workflow and additional content are available on the Biosecurity Commons YouTube channel.
• For more detail on the underlying methodology, see Camac et al. (2020, 2021, 2024) and related references listed below.
• The  Impact Analysis workflow  explains how establishment likelihood maps can be combined with impact data to quantify likely harm.

Biosecurity Commons is designed for applied biosecurity practitioners who need defensible, reproducible risk maps but may not have time for extensive coding or GIS work.

• Accessible analytics: cloud-based workflows mean you do not need to install software or manage compute resources. You can run complex analyses through a web interface.
• Data at your fingertips: the platform connects to thousands of trusted datasets (species records, traits, environmental layers, climate projections and more).
• Reproducible and shareable: workflows and outputs can be documented, rerun and shared among agencies and experts, supporting nationally consistent assessments.
• Integrated with other workflows: risk maps produced on the platform can be directly used in dispersal modelling, surveillance design and proof-of-freedom analyses.

An establishment likelihood map is a spatial layer where each cell stores the combined probability that a threat will establish at that location. It is derived by combining:

• Abiotic suitability (for example, climate and environmental conditions).
• Biotic suitability (for example, host or habitat availability).
• Propagule pressure (for example, volume and distribution of threat arrivals).

The map shows relative risk across the landscape – higher values indicate locations where the conditions and arrivals suggest establishment is more likely. For more context, see Establishment likelihood and its components.

Abiotic suitability describes how suitable the non-living environment is for a threat. This typically includes climate (temperature, rainfall, humidity, seasonality), but can also include soil, elevation and disturbance regimes.

On Biosecurity Commons, abiotic suitability is often estimated using species distribution models that relate known occurrences (or expert-mapped distributions) to environmental variables.

Practical tips:

• Use environmental covariates that are biologically meaningful for the threat (for example, maximum temperature of the warmest month, moisture indices, frost days).
• Consider model transferability – especially when projecting suitability under new climates or into regions where the species is not yet present.
• Treat marginally suitable areas cautiously: do not assume that low suitability is equivalent to zero risk, especially for highly adaptable species.

Biotic suitability summarises how suitable the living environment is for a threat. It typically depends on the distribution and density of hosts, suitable habitat or key resources, and sometimes on competitors, natural enemies or mutualists.

In practice, biotic suitability is often approximated using:

• Land use and land cover layers (for example, crop type, forest type, urban areas).
• Vegetation maps or host distribution datasets.
• Composite indices that combine several habitat indicators.

Where host or habitat data are limited, expert judgement and simple proxies (for example, “all irrigated agriculture” or “all broadleaf forest”) can be used, with uncertainty documented in the model description.

Propagule pressure refers to how often and how many viable threats arrive at different locations via relevant pathways. For modern biosecurity threats, this is often driven by the movement of people and goods across borders and within countries.

To estimate propagule pressure for a given threat, you might combine:

• Border interception or contamination data.
• Trade and transport volumes (for example, imports, shipping routes, air traffic).
• Post-border movement patterns (for example, road networks, freight hubs, distribution centres).
• Expert elicitation where empirical data are sparse.

In Biosecurity Commons, these data can be translated into spatial layers that describe relative arrival likelihood across the landscape.

Beyond the core abiotic, biotic and propagule pressure components, it is important to consider:

• Time horizon: are you mapping current risk, near-term scenarios, or long-term climate change projections?
• Spatial resolution: is the grid size appropriate to the biology of the threat and the management decisions you want to support?
• Data quality and uncertainty: where are the major gaps in data, and how might they bias the map?
• Pathway coverage: have you captured all important pathways, not just the most obvious ones?
• Assumptions and simplifications: clearly document any inputs you have omitted (such as a uniform abiotic suitability for highly climate-tolerant stored-product pests).

Once a risk map has been drafted, there are several levers you can use to refine and improve it:

• Refine inputs: update abiotic layers, host data or pathway assumptions with more recent or higher-resolution datasets.
• Stress-test assumptions: run sensitivity analyses by varying key parameters, such as the strength of pathway links or the importance of particular habitat types.
• Engage experts: share preliminary maps with subject-matter experts and capture their feedback on unlikely “hotspots” or missing pathways.
• Calibrate against known incursions: where possible, compare the map with historic establishment records to see whether the model captures known risk patterns.

Establishment likelihood maps are powerful, but they remain simplifications of reality. Key limitations include:

• Data limitations: occurrence, host and pathway data can be incomplete or biased, particularly for emerging threats.
• Model assumptions: the multiplicative structure assumes independence between components and may not capture all interactions (for example, climate–land use interactions).
• Static representation: maps usually represent a snapshot in time, whereas many threats and pathways are dynamic.
• Uncertainty communication: maps often show a single “best” estimate, and it can be challenging to communicate the full uncertainty to decision-makers.

These limitations mean that establishment likelihood maps should be viewed as structured decision support tools, not as perfect predictions.

Establishment likelihood maps are best interpreted as relative risk surfaces: locations with higher values are more likely to support establishment than locations with lower values, given the assumptions built into the model.

When interpreting maps:

• Focus on broad patterns (for example, hotspots, gradients) rather than individual grid cells.
• Consider the intended decision context: for early detection surveillance, high-risk “frontier” areas may be more important than already-established cores.
• Check that the map is consistent with expert expectations and known biology of the threat.

The core mathematical idea is that establishment at a location requires three things to occur: abiotic suitability, biotic suitability and propagule pressure (threat arrivals). In Biosecurity Commons, these are represented as probabilities (or probability-like indices) between 0 and 1 and combined multiplicatively:

A more detailed description of this formulation – including the definition of events, assumptions, weighting and scaling – is provided in the section Mathematics of establishment likelihood above.

• Camac, J.S., et al. (2020). Risk mapping and surveillance design frameworks. Centre of Excellence for Biosecurity Risk Analysis (CEBRA) Report.
• Camac, J.S., et al. (2021). Establishment likelihood and proof-of-freedom analytics. CEBRA Report.
• Camac, J.S., et al. (2024). Invasion risk and establishment frameworks. Open access monograph.
• Dodd, A.J., et al. (2016). Biological invasions and propagule pressure: insights for risk assessment. Biological Invasions.
• Venette, R.C., et al. (2010). Pest risk maps for invasive alien species: a road map for improvement. Bioscience.