TABLE OF CONTENTS

 


Risk Mapping


Risk Mapping estimates where a threat might arrive and establish. Understanding where a harmful invasive species might arrive is critical for biosecurity decision support, especially given the finite resources available for surveillance.  

This explainer focuses on estimating the likelihood of a biosecurity threat establishing and sets out to answer frequently asked questions on risk mapping.


Other resources


Frequently asked questions


Why build a risk map on Biosecurity Commons?

 

Accessible risk mapping tailored to applied biosecurity practitioners

The risk mapping workflow on Biosecurity Commons is meticulously designed for applied biosecurity practitioners, providing effortless access to advanced risk mapping analytics and an extensive array of spatial datasets. Tailored for accessibility, it empowers individuals with limited geospatial or statistical training to leverage thousands of datasets, powerful computational resources, and state-of-the-art risk mapping tools, enabling them to make informed decisions and effectively communicate biosecurity risks.

Access to thousands of trusted spatial datasets

The platform provides easy access to thousands of trusted species occurrence records, species traits records, environmental and climate projection layers.

Everything you need on the cloud

There is no need to install anything and direct access to NECTAR cloud resources means you can perform complex analyses without burdening your computer and you can run workflows without coding experience.

Reproducible and shareable: Enhancing collaboration and expertise exchange for improved model development 

Risk maps generated in Biosecurity Commons are standardised, reproducible, and sharable among organisations and experts. This capability fosters robust collaboration and knowledge exchange across both government and non-government entities. As a result, it enhances model development and paves the way for nationally endorsed risk analytics that empower all stakeholders to effectively coordinate and elevate their efforts in managing biosecurity risks.

Enhancing biosecurity decision-making
Risk maps are essential decision-support tools for government and industry. These maps serve multiple critical functions:

  • Prioritising surveillance efforts: Risk maps direct monitoring resources to areas with the highest likelihood of biosecurity threat establishment, enhancing early detection capabilities.
  • Assessing surveillance coverage: Risk maps evaluate the effectiveness of surveillance activities in covering high-risk regions.
  • Estimating area freedom: Risk maps facilitate trade and market access by quantifying the probability of a region being threat-free, based on surveillance data, detection probabilities, and establishment likelihood maps.
  • Spread modelling: Risk maps enhance the realism of spread models by identifying potential initial establishment sites based on pathway propagule pressure and biotic/abiotic suitability, while also informing constraints on subsequent spread events.
  • Impact analysis: Risk maps are an integral component of models used to assess the potential harm a threat could inflict on economic, environmental, and societal values.


Transparent, evidence-based and easily integrated into other workflows

Biosecurity Commons fosters transparency by providing a centralized platform for risk assessment data. This supports evidence-based policy development and ensures stakeholders across government, industry, and research can collaborate effectively.

By leveraging Biosecurity Commons for risk mapping, biosecurity professionals gain a powerful, data-driven tool to prioritise threats, refine strategies through collaboration, and enhance national biosecurity preparedness and response. 

Risk maps developed on the platform can be used directly as inputs in many other workflows, such as Dispersal Modelling, Surveillance Design and Proof of Freedom. You can find out more about this in our Risk Mapping quick start guide.

Key sources for more information: (Camac et al. 2020, Camac et al. 2021,  Camac et al. 2024, Dodd et al. 2016, Venette et al. 2010)

 

What is establishment likelihood and how is it calculated?


Establishment likelihood, or establishment potential, is the probability that an exotic threat will become established in a particular location. It is governed by three fundamental spatial constraints: propagule pressure, abiotic suitability, and biotic suitability (Figure 1).

Figure 1: The three main elements governing the likelihood of establishment of exotic species in the introduced region (Derived from Camac et al. 2024)

  • Abiotic suitability: refers to the suitability of the non-living environment for a biosecurity threat. This includes factors like climate (too hot, too dry), terrain, or disturbance regimes. Key Consideration - Marginally suitable areas should not be ruled out prematurely, as species may adapt.
  • Biotic suitability: refers to the suitability of the living environment for a threat. This includes factors like the presence of suitable habitat, food sources, hosts, or other species that could facilitate or hinder establishment.  Key Consideration - Limited distributional data on hosts or empirical data on threat-host relationships may introduce uncertainty.
  • Propagule pressure: also known as introduction rate or contamination rate, is a measure of the likelihood that a threat will arrive at a location of interest based on different pathways of entry and movement. Key Consideration - Arrival probabilities should account for all pathways that may allow a threat to be introduced to a region of interest, whether they be human-mediated (trade, human movement) and/or natural (e.g. wind).

To calculate the establishment likelihood, the probabilities of each of these three barriers are multiplied together. For example:

Establishment Likelihood = Pr(Abiotic Suitability) x Pr(Biotic Suitability) x Pr(Propagule Pressure)

Figure 2: The product of the three geographic barriers to oriental fruit fly establishment can be used to approximate establishment potential (likelihood of establishment). 

This principle is based on the idea that if any of the three barriers to establishment is not overcome (i.e. the probability is zero), then establishment cannot occur [2]. The resulting grid includes values in each grid cell that represent the combined likelihood of establishment in that specific location.

The standard risk mapping approach on Biosecurity Commons allows users to:

  • Estimate the probability that the abiotic (e.g. climatic) environment is suitable – often estimated using species distribution models
  • Estimate the probability that the biotic (e.g. habitat) environment is suitable – often estimated using landuse and vegetation spatial datasets
  • Estimate the likelihood a viable threat will arrive at a given location – often estimated using a combination of expert elicitation, border surveillance data and proxies for likely post-border movement
  • Combine these three likelihoods to derive an estimate of establishment likelihood

Find out more about the maths used to calculate establishment likelihood. 

          

What are the recommendations for estimating abiotic suitability?


Abiotic barriers that limit a species' potential distribution vary widely and include factors like climate, disturbance patterns, and terrain features. At large scales, such as global or continental levels, climate is believed to play a key role in determining a species' distribution (Araújo and Rozenfeld 2014). Over the past few decades, global databases of climate and biological data have emerged, along with a variety of statistical and mechanistic species distribution models (SDMs). In invasive species management, SDMs often use global climate data—easily accessible environmental information—leading to their designation as "climate suitability models." Despite the many SDM methods available and differing opinions on their application, no single approach has been proven to be the best for predicting the potential distribution of invasive species (Camac et al. 2024).

Biosecurity Commons provides users with a wide range of statistical SDM functionality that allows users to generate maps identifying where the environment may be suitable for a species based on climatic and/or habitat preferences.

Biosecurity Commons integrated models

Biosecurity Commons provides direct integrated support for two profile (i.e. presence-only) models: Range-bagging and Climatch SDM algorithms commonly used for invasive species. These algorithms have seen recent applications to invasion biology and appear promising in the context of biosecurity. Part of their appeal is that no absences or background data are required – presence data are sufficient (Camac et al.,2020, Hill et al., 2022, Camac et al. 2024). This removes several subjective decisions required in the modelling process and focuses solely on the data commonly available for most threats – presence locations. Moreover, using Biosecurity Commons integrated models provides users with on-platform data cleaning routines powered by CoordinateCleaner (Zizka et al. 2019) – functionality not currently available on EcoCommons (see below). 

EcoCommons models

If users wish to utilise other SDM algorithms, they can directly access functionality available on our sister platform, EcoCommons. EcoCommons, a SDM specific platform, provides users with an additional 16 SDM algorithms spanning profile models (e.g.BIOCLIM), machine learning models (e.g. MaxEnt), statistical models (e.g. Generalised Additive Models) and geographic models (e.g. Convex and Voronoi hulls). While EcoCommons does not currently offer the data cleaning routines available on Biosecurity Commons, it does provide functionality for forecasting species distributions under climate change.

What model should I use?

If you are unsure which model to use, and you are uncertain which climatic predictors should be included, we generally recommend using Biosecurity Common’s Range Bagging algorithm (Drake 2015). Range bagging is an algorithm that estimates the environmental limits of a species’ habitat by calculating convex hulls around environmental conditions at occurrence locations. Environments that fall within the hull are defined as suitable, whereas those that fall outside are considered unsuitable. This process is then repeated using random subsets of both occurrence records as well as available environmental covariates (e.g., annual rainfall, mean annual temperature, etc.). 

This approach has seen recent applications to invasion biology and appears promising in the context of biosecurity. Part of the appeal is that:

  • Works well with presence-only data: Range-bagging effectively utilises presence-only data, which is often the only type of data available for biosecurity environmental space threat species, especially for those not yet established in a region. By focusing on environmental limits and providing interpretable, comparable results (the proportion of replicate hulls in environmental space that contain given environmental conditions) the method is well-suited to identifying areas of potential establishment.
  • Fewer subjective modelling decisions: Because range bagging is a true “presence-only” model, users do not need to define the number, extent and distribution of background/pseudo absence points required in other commonly used algorithms such as Maxent. This in turn removes reduces the number of subjective decisions the modeller is required to make.
  • Simple to implement and performs well: Range-bagging produces relatively good predictive performance that transfer well to other regions. It is also less computationally expensive and has little need for manual tweaking or parameter fitting, so it is efficient to run.
  • Reduces overfitting: By creating many simple models based on different subsets of environmental variables, Range-bagging reduces the risk of overfitting to specific environmental conditions. In other words, results will tend to be more conservative, avoiding incorrectly removing areas from the potential distribution and indirectly this can help correct for sampling bias. Work by Breiner et al. (2017) has found that using ensembles of small models, each with only two variables, often outperforms standard SDM methods.
  • Not relying on a single model: Range-bagging works by ensembling hundreds or thousands of competing models. As such it allows for uncertainty among models as well as uncertainty in covariate selection.
  • Easy to interpret: The output of a range bag model details the proportion of models ensembled that indicate a given location is suitable. For example, a suitability score of 0.1 would indicate only 10% of the estimated convex hulls ensembled deemed that location suitable. By contrast, a score of 0.9 would indicate that 90% of estimated convex hulls deemed that location climatically suitable.
  • Outputs are comparable across threats: For presence-background models like Maxent, scores are not directly comparable between species. For example, a score of 0.6 for one species doesn't mean the same for another. This is because these models can't accurately estimate how common occurrences are compared to absences (i.e., the species' prevalence). Although we call these scores "relative suitability scores," this can be misleading, as it might imply that a score of 0.7 is twice as suitable as 0.35. Instead, Maxent scores should only be seen as rankings of habitat suitability. In contrast, range bagging is a presence-only model that doesn't need information about prevalence. Its scores are comparable across species. A score of 0.6 in a range bag model means that 60% of the ensembled convex hulls identify a location as suitable, and this interpretation holds true for different species.
  • Direct access to occurrence record cleaning functionality: As Range bagging is a “Biosecurity Commons Integrated Algorithm” users can directly use it with the platforms SDM workflow. This means that users can also clean their occurrence records by using standard rules available in CoordinateCleaner.

Regardless of the method selected there are a wide variety of climate and environmental variables (i.e. WorldClim, CliMond, CHELSA) to use within the Biosecurity Commons platform, often with different spatial resolution and time frames.

Key sources for more information: (Camac et al. 2020, Camac et al. 2021,  Camac et al. 2024, Catford et al. 2009,Drake 2015, Elith 2017, Elith et al. 2010, Fourcade et al. 2017, Hill et al. 2017, Phillips et al. 2009, Renner et al. 2015, Syfert et al. 2013, Warton and Sheperd 2010)

 

What are the recommendations for estimating biotic suitability?


There are cases where mapping biotic suitability is very straightforward. For example, if an insect pest shows an extremely strong preference for one group of plant species, mapping biotic suitability only requires good mapping of where that group of species occurs, i.e. clover, or citrus. When specific knowledge of where specific hosts are found is lacking, proxies can be used to capture all the areas they might occur. 

General Recommendations:

  • Consider the threat's host range: Start by understanding the known host range of the threat. This information can be obtained from literature reviews, expert knowledge, or online databases.
  • Identify suitable habitat: Consider the habitat requirements of the threat, such as vegetation type, land use, and proximity to water sources.   
  • Use high-resolution data: When available, use high-resolution spatial data to capture the fine-scale distribution of suitable host material.  
  • Integrate multiple data layers: Combine different data layers, such as land use, vegetation, and host plant distribution, to create a more comprehensive picture of biotic suitability.  
  • Account for uncertainty: Recognize that there is always uncertainty in estimating biotic suitability. Use conservative assumptions and consider a range of possible scenarios i.e. there may be hosts in Australia that have never been encountered by the threat but are none-the-less suitable.

The following data layers are commonly used in Australia to approximate the geographic distribution of suitable host material.

  • Australian Land Use and Management Classification (ALUMC): This comprehensive dataset classifies the dominant land use of each 50m x 50m grid cell across Australia, providing detailed information on agricultural production, natural areas, and urban areas. 
  • National Vegetation Information System (NVIS): This 100m resolution raster classifies vegetation types based on structure and dominant species, helping to refine the identification of suitable habitats within broader land use categories. 
  • Fractional Cover: This enables measurement of green (leaves, grass, and growing crops), brown (branches, dry grass or hay, and dead leaf litter), and bare ground (soil or rock) in any area of Australia at any time since 1987. It can be used to identify areas with sufficient vegetation to support certain weeds, pests or diseases.
  • Species-specific host distribution data: This might include the geographic distribution of specific host plants or habitat types required by a particular invasive species. This data can be sourced from various sources, including expert knowledge, field surveys, an SDM of the host, or online databases like GBIF. 

Key sources for more information: (Camac et al. 2020, Camac et al. 2021,  Camac et al. 2024, Catford et al. 2009)

Note: If the abiotic or biotic suitability is expected to be uniform across space then it can be ignored. For example the Khapra beetle is a threat of stored food stuff. It’s not typically subjected to ambient climatic conditions as it often occurs in pantries or food storage facilities. It is also very hardy to a wide range of climate conditions and if conditions are unfavourable it can go into diapause for seven years. For this threat the abiotic suitability can be ignored and it is better to focus on biotic suitability and arrivals. If not required, Biosecurity Commons allows users to easily remove this input by simply selecting the “abiotic” branch and selecting “remove input” .

 

What are the recommendations for estimating propagule pressure (threat arrivals)?


Most contemporary introductions of exotic threats are driven by the movement of people and goods across borders (Hulme 2021). To assess the likelihood of an exotic threat arriving at a location, biosecurity practitioners need to consider both the chance of viable threats bypassing border controls and how those threats might disperse after entry.

1. Estimating Leakage and Viability

The likelihood that a threat enters a country undetected, termed leakage, and remains viable can be estimated using a combination of:

  • Border screening and interception data provide records of detected threats. High interception rates may indicate effective screening or high threat volumes, while low detection rates in known high-risk pathways may suggest leakage, i.e., undetected threats passing through. Additionally, the condition of intercepted organisms (e.g., alive, reproductive stage) can inform estimates of viability, indicating whether threats arriving at the border can establish.
  • Pathway analysis evaluates the volume and type of carriers (e.g., cargo, mail, passengers) arriving from regions where the threat is present (Tingley et al. 2018). This helps estimate the likelihood of introduction by identifying which routes are most likely to carry viable threats, how frequently they occur, and what types of commodities or movement patterns are involved. Combined with historical interception rates, this analysis can suggest where leakage is most likely to occur and what types of threats are most viable upon arrival. 
  • Expert elicitation helps estimate viability when empirical data is limited or absent, especially for emerging threats. Structured consultation with specialists are generally used to form these estimates.

These sources can be used to define probability bounds for both leakage (i.e., failure of detection) and the likelihood that the threat remains viable upon arrival.

2. Dispersing Arrival Risk

Once leakage and viability estimates are made, they must be coupled with a spatial model of post-border dispersal. This involves estimating where threats are likely to go after crossing the border.

Entry pathways are often tied to human behaviour and infrastructure. For example:

  • International tourists cluster in high-tourism areas.
  • Returning residents or postal deliveries tend to follow population density.
  • Imported goods are usually destined for major cities or distributed based on land use (e.g., agriculture, industrial zones).

Even in the absence of detailed post-border movement data, dispersal patterns can often be inferred using variables that are expected to be correlated with human movement, such as:

  • Population density
  • Road and freight infrastructure
  • Land use and agricultural intensity

These proxies align with studies showing strong spatial correlations between these variables and initial detections of exotic species (e.g., Dodd et al. 2016).

Biosecurity Commons provides users with generic functions to develop their own weighting layers for arrival likelihoods from different pathways, and for Australian case studies, we also have pre-defined pathway layers derived directly from Camac et al. 2021.

A screenshot of a computer AI-generated content may be incorrect.

Figure 3: A screenshot from the Biosecurity Commons platform showing the available threat arrival layers

Useful data to inform arrival risk:

Variables that users need to supply

  • Number of Vessels and containers arriving at ports or points of interest. 
  • Location of arrival hubs. This can include the mapping of nurseries, farms, airports etc and some assumptions about how far threats might spread from those locations.
  • Border interception data: Empirical data on interceptions of pests and diseases at the border can inform pathway analyses and help estimate leakage rates (the number of biosecurity incursions that get past the border per time-period). Camac et al. 2024.  

These can be estimated using data available on the platform

  • Human population density, Human Pathway: Often used as a proxy for the distribution of propagule pressure from pathways like mail and returning residents. The ABS has gridded data (rasters) on population density.
  • Tourist accommodation density, Tourist Pathway: This data is used in conjunction with distance from international airports to approximate the distribution of propagule pressure from international tourists. 
  • NRM usage statistics, Agriculture Pathway: These statistics can be used to distribute the risk from imported fertilizer based on the regions where it is used. 
  • Wind data / Wind pathway: Daily wind simulations, particularly for winds originating from Papua New Guinea, Indonesia, Pacific Islands, and New Zealand, are used to estimate the frequency of potentially threat-carrying wind events crossing Australian coasts. 

Key sources for more information: (Camac et al. 2020, Camac et al. 2021,  Camac et al. 2024, Catford et al. 2009, to elicit needed experts from experts see Hemming et al. 2018, Hulme 2021, Venette et al. 2010)

 

 

What additional things should be considered while making a risk map?


A diagram of steps to estimating establishment Description automatically generated

Figure 4: Steps to estimating establishment likelihoods on Biosecurity Commons

 

A literature review can: 

  • Reveal existing risk maps or data that can be incorporated into the risk map 
  • Inform on the threat's biology, such as its physiological tolerances Bennett et al. 2018 to different climate conditions Kearney et al. 2008 and its host preferences
  • Identify most probable points and pathways of entry

Collating and cleaning data can: 

  • Collating existing data can improve the precision of estimates and there are growing volumes of data available on Biosecurity Commons that can be used and shared with those people you’re working with
  • Cleaning data can have massive impacts on the accuracy and precision of estimates, and includes removing errors in taxonomy Chamberlain and Szocs 2013, location Zizka et al. 2019, location names Boyle et al. 2022, time and removal or irrelevant records

Selecting the extent and resolution can:

  • Ensure you are creating a map that covers the areas where resource allocation decisions are being made, while ensuring the information within the boundary selected is at the same quality
  • Align the resolution to the scale at which surveillance and management efforts will be applied while keeping the resolution as fine as possible given the available data.
  • Some errors introduced to spatial locations can be introduced when reprojecting data from one coordinate reference system (CRS) to another. Use the CRS for the data that is most important for the risk mapping result to align all other spatial data. In many biosecurity applications, the use of an equal area projection is appropriate (i.e. EPSG:3577, EPSG:9473). This ensures that the spatial distribution of risk is proportional to the actual area being represented, allowing for more accurate comparisons and assessments of risk across different regions. This is particularly important in risk analysis, as it helps prevent misinterpretations that could arise from distortions in area size, leading to better-informed decision-making and resource allocation. especially when making decisions about where finite surveillance or control resources are allocated.

Validate and interpret your results:

  • Document data and assumptions made, then share with other experts and practitioners to improve the components of the map
  • Avoid thresholding (i. e. converting continuous values with more information into a set of 1s and zeros)
  • Validate your map as often as possible, by sharing it with others, plotting new detections on the map etc. Remember results are relative likelihoods, so results might not be directly comparable between risk maps

Key sources for more information: (Bennett et al. 2018, Boyle et al. 2022, Camac et al. 2020, Camac et al. 2021, Camac et al. 2024, Catford et al. 2009, Chamberlain and Szocs 2013, Kearney et al. 2008, Lahoz-Monfort et al. 2014, Venette et al. 2010, Zizka et al. 2019)

 

          

Once a risk map is drafted what are some potential areas to focus on to improve results?

 

Enhance data quality and validation: Ensure that species occurrence data, environmental variables, and pathway data are accurate and up to date. Use standardized tools to clean, validate, and refine datasets, removing errors and biases that may distort predictions. 

 

Incorporate uncertainty and sensitivity analysis: Identify sources of uncertainty in the model arising from incomplete data, assumptions about species tolerances, and environmental variability. Sensitivity analyses can help determine which parameters have the greatest influence, allowing for more informed refinements (see how results change when parameter estimates and assumptions are adjusted). Critically examine the assumptions made in propagule pressure estimates. Explore utility of higher resolution biotic suitability datasets or species-specific host distribution data and consider adding host susceptibility or natural threat enemies to the mapping workflow.

 

Refine model assumptions and parameters: Review the thresholds and assumptions applied during model development. Avoid overly restrictive assumptions about species distributions, and consider how additional factors like biotic interactions, land use changes, or climate change projections may alter establishment risk.

 

Integrate field data and expert feedback: Collaborate with on-ground practitioners to compare model predictions with real-world observations. Field validation and expert review can help adjust risk estimates and ensure the map reflects practical, on-the-ground conditions.

 

Improve communication and decision support: Ensure that risk maps are clear, interpretable, and actionable for decision-makers. Provide explanations of model limitations, highlight areas of high uncertainty, and integrate visualisation tools to support effective resource allocation and biosecurity planning.


 

What are some of the limitations of establishment likelihood maps?


Risk maps are only as reliable as the data used to create them. Limited species occurrence records outdated environmental data, and incorrect assumptions about species tolerances or arrival pathways can all lead to misleading predictions.

Risk maps tend not to include complex biotic interactions. Factors like predation, competition or dynamic host availability are rarely included in a risk map, but such considerations may be important in some use cases, if hard to model. 

Risk maps are usually static in space and time. A risk map is a snapshot that may miss seasonal variations in threat life cycles, temporal changes in climate, land-use or emerging pathways. 

Use of a risk map needs to match the spatial scale at which it was developed. A model designed at a continental scale may not accurately reflect local establishment risks, leading to misinformed management decisions if applied at the wrong resolution.

Interpret risk maps based on underlying assumptions and uncertainties. Users should document underlying assumptions, and limitations in the data and then clearly communicate those caveats and limitations to those looking to use a risk map which is always simplifying complex processes. The overall error in the establishment likelihood estimate is a combination of the errors in each variable, so high uncertainty in one estimate carries through to the final estimate. 


How do I interpret a risk map?


A risk map shows the relative (conditional) likelihood that a viable threat (e.g., pest or disease) could arrive and establish at a particular location. It combines two key components:

  • Arrival Likelihood – how likely it is that a viable threat reaches the location via human or goods movement (propagule pressure).
  • Pest Suitability – Does the location have the right climate (abiotic) and ecological (biotic) conditions for the threat to survive and reproduce.

This map helps biosecurity practitioners prioritise where to focus surveillance and control, based on where threats are most likely to arrive and succeed.


What is the maths behind calculating establishment likelihood?


Mathematically, what happens is that user defines the lower and upper confidence limits of average annual leakage rate ( ) for pathway, k. These confidence limits are then used to estimate the mean and standard deviation of a log-normal distribution, from which the rate parameter λk can be estimated: 

(1)

(2)

(3)

 

The discrete expected number of entry events, n, in a given year, can then be simulated from a Poisson distribution using the rate parameter, λ:

(4)

However, just because a leakage event may occur, does not mean it is viable for establishment. Different pathways of entry are likely to have varying survival rates. Moreover, some pathways may allow for larger founder population sizes to occur. Biosecurity Commons allows users to account for these differences explicitly by specifying the lower and upper probability bounds, ( ), for pathway k. These lower and upper bounds should explicitly consider both the joint probability that 1) the threat of interest can survive the pathway transportation; and 2) whether a leakage event would likely contain a population size that is conducive of establishment. These viability bounds are used to estimate the mean and standard deviation of a logit-normal distribution, from which the likelihood of establishment viability, pk, can be simulated: 

(5)

(6)

 

(7)

(8)

Biosecurity Commons then uses the simulated distributions of and to estimate the expected number of viable incursion events for each pathway, k, using a Binomial distribution: 

(9)

is then used to create a discrete probability mass function, , that defines the discrete likelihood that N viable incursions will occur in a given year, attributable to pathway k. 

Biosecurity Commons then uses to estimate the probability raster cell i receives one or more viable incursions from pathway k, as: 

(10)

where is a pathway-specific weight that estimates the likelihood that a pathway unit for pathway k will arrive in cell i. For example, for the mail pathway, the geographic distribution of mail items is assumed to follow that of human population density, hence for that pathway is defined as the population density at cell i, divided by the sum of population density for the entire study area (Australia). The complement, 1 − , seen in Equation 2.13, gives the likelihood that a given pathway unit for pathway k does not arrive in cell i. 

The probability of one or more viable incursions at cell i across pathways, then becomes: 

(11)

edmaps then estimates the probability of establishment by weighting by the estimated biotic and abiotic suitability of the location: 

(12)

Key sources for more information: (Camac et al. 2020, Camac et al. 2021,  Camac et al. 2024, Dodd et al. 2016)

 

References


Bennett, J. M., P. Calosi, S. Clusella-Trullas, B. Martínez, J. Sunday, A. C. Algar, M. B. Araújo, B. A. Hawkins, S. Keith, I. Kühn, C. Rahbek, L. Rodríguez, A. Singer, F. Villalobos, M. Ángel Olalla-Tárraga, and I. Morales-Castilla. 2018. “GlobTherm, a Global Database on Thermal Tolerances for Aquatic and Terrestrial Organisms.” Scientific Data 5 (1):180022. https://doi.org/10.1038/sdata.2018.22.

Boyle, B. L., B. S. Maitner, G. G. C. Barbosa, R. K. Sajja, X. Feng, C. Merow, E. A. Newman, D. S. Park, P. R. Roehrdanz, and B. J. Enquist. 2022. “Geographic Name Resolution Service: A Tool for the Standardization and Indexing of World Political Division Names, With Applications to Species Distribution Modeling.” https://doi.org/10.1101/2022.04.25.489424

Camac, J. S., Palma, E., & Baumgartner, J. B. (2024). Map: Creating Maps of Establishment Potential. In Biosecurity. Taylor & Francis.  

Camac, J. S., Baumgartner, J. B., Garms, B., Robinson, A., & Kompas, T. (2021). Estimating trading partner exposure risk to new pests or diseases. Technical Report for CEBRA project 190606. Centre of Excellence for Biosecurity Risk Analysis, The University of Melbourne.

Camac, J. S., Baumgartner, J. B., Hester, S., Subasinghe, R., & Collins, S. (2021). Using edmaps & Zonation to inform multi-pest early-detection surveillance designs. Technical Report for CEBRA project 20121001. Centre of Excellence for Biosecurity Risk Analysis, The University of Melbourne.  

Camac, J. S., Baumgartner, J., Robinson, A., Elith J (2020) Developing pragmatic maps of establishment likelihood for plant pests. Technical Report for CEBRA project 170607.

Catford, J. A., Jansson, R., & Nilsson, C. (2009). Reducing redundancy in invasion ecology by integrating hypotheses into a single theoretical framework. Diversity and Distributions, 15, 22–40. https://doi.org/10.1111/j.1472-4642.2008.00521.x  

Chamberlain, S., & Szöcs, E. (2013). Taxize: Taxonomic Search and Retrieval in R [version 2; Peer Review: 3 Approved]. F1000Research, 2(191). https://doi.org/10.12688/f1000research.2-191.v2

Dodd, A. J., McCarthy, M. A., Ainsworth, N., & Burgman, M. A. (2016). Identifying Hotspots of Alien Plant Naturalisation in Australia: Approaches and Predictions. Biological Invasions, 18(3), 631–645. https://doi.org/10.1007/s10530-015-1035-8

Drake, J. M. (2015). Range Bagging: A New Method for Ecological Niche Modelling from Presence-Only Data. Journal of the Royal Society Interface, 12(107), 2015008     https://doi.org/10.1098/rsif.2015.0086

Elith, J. 2017. “Predicting Distributions of Invasive Species.” In Invasive Species: Risk Assessment and Management, edited by Andrew P. Robinson, Mark A. Burgman, Mike Nunn and Terry Walshe, 93–129. Cambridge: Cambridge University Press.

Elith, J., M. Kearney, and S. Phillips. 2010. “The Art of Modelling Range-Shifting Species.” Methods in Ecology and Evolution 1 (4):330–342. https://doi.org/10.1111/j.2041-210X.2010.00036.x

Fourcade, Y., Besnard, A. G., & Secondi, J. (2017). Paintings predict the distribution of species, or the challenge of selecting environmental predictors and evaluation statistics. Global Ecology and Biogeography, 27, 245–256. https://doi.org/10.1111/geb.12684  

Hill, M. P., Gallardo, B., & Terblanche, J. S. (2017). A global assessment of climatic niche shifts and human influence in insect invasions. Global Ecology and Biogeography, 26, 679–689. https://doi.org/10.1111/geb.12578

Hemming, V., Burgman, M. A., Hanea, A. M., McBride, M. F., & Wintle, B. C. (2018). A practical guide to structured expert elicitation using the IDEA protocol. Methods in Ecology and Evolution, 9(1), 169-180. https://doi.org/10.1111/2041-210X.12857

Hulme, P. E. 2021. “Unwelcome Exchange: International Trade as a Direct and Indirect Driver of Biological Invasions Worldwide.” One Earth 4 (5):666–679. https://doi.org/10.1016/j.oneear.2021.04.015.

Kearney, M., B. L. Phillips, C. R. Tracy, K. A. Christian, G. Betts, and W. P. Porter. 2008. “Modelling Species Distributions Without Using Species Distributions: the Cane Toad in Australia Under Current and Future Climates.” Ecography 31 (4):423–434. https://doi.org/10.1111/j.0906-7590.2008.05457.x.

Lahoz-Monfort, J. J., G. Guillera-Arroita, and B. A. Wintle. 2014. “Imperfect Detection Impacts the Performance of Species Distribution Models.” Global Ecology and Biogeography 23 (4):504–515. https://doi.org/10.1111/geb.12138.

Phillips, S. J., Dudík, M., Elith, J., Graham, C. H., Lehmann, A., Leathwick, J., & Ferrier, S. (2009). Sample selection bias and presence-only distribution models: implications for background and pseudo-absence data. Ecological Applications, 19(1), 181-197. https://doi.org/10.1890/07-2153.1

Renner, I. W., Elith, J., Baddeley, A., Fithian, W., Hastie, T., Phillips, S. J., Popovic, G., & Warton, D. I. (2015). Point process models for presence-only analysis. Methods in Ecology and Evolution, 6(4), 366–379. https://doi.org/10.1111/2041-210X.12352    

Syfert, M. M., Smith, M. J., & Coomes, D. A. (2013). The Effects of Sampling Bias and Model Complexity on the Predictive Performance of MaxEnt Species Distribution Models. PLoS ONE, 8(2), e55158. https://doi.org/10.1371/journal.pone.0055158  

Venette, R. C., Kriticos, D. J., Magarey, R. D., Koch, F. H., Baker, R. H. A., Worner, S. P., Raboteaux, N. N. G., McKenney, D. W., Dobesberger, E. J., Yemshanov,D., De Barro, P. J., Hutchison, W. D., Fowler, G., Kalaris, T. M., Pedlar, J., (2010) Pest Risk Maps for Invasive Alien Species: A Roadmap for Improvement, BioScience, Volume 60, Issue 5, Pages 349–362,  https://doi.org/10.1525/bio.2010.60.5.5   

Warton, D. I., & Shepherd, L. C. (2010). Poisson point process models solve the" pseudo-absence problem" for presence-only data in ecology. The Annals of Applied Statistics, 1383-1402.

Zizka, A., Silvestro, D., Andermann, T., Azevedo, J., Ritter, C. D., Edler, D., ... & Antonelli, A. (2019). CoordinateCleaner: Standardized cleaning of occurrence records from biological collection databases. Methods in Ecology and Evolution, 10(5), 744–751. https://doi.org/10.1111/2041-210X.13152