What is Surveillance Design? 


At Biosecurity Commons the Surveillance Design workflow is built to help design adequate and cost-effective monitoring strategies for invasive species which have already or might have crossed through border controls. There are limited resources to conduct surveillance, so it is critical to optimise how available resources are deployed to get the most out of each deployment. Effective surveillance design is critical to detect an invasive species incursion early enough to allow response measures to be taken to minimize impacts. Surveillance systems can also be employed to define the boundary of an already established incursion or to track the effectiveness of response, containment, or control measures. Overall, surveillance design optimises available monitoring efforts to track the status of invasive species in selected areas. This tracking of status is critical to protect the health and productivity of Australian ecosystems as well as human health and a variety of industries. 


The Biosecurity Platform can be tailored to address your surveillance question.   Please get in touch if you would like to work on a surveillance question with our team.




Figure 1: Post-border surveillance and incursion management responses with an example scenario (adapted from Hester et al. 2010). 



Our surveillance design workflow is built on the foundation of mathematical optimization which is a process where the best value is selected from a set of alternative values based on a specific criterion. The objective function in an optimisation problem measures the quality of a solution, and it is optimized when a global minimum or maximum is identified.  The solution involves systematically selecting input values from an allowed subset of possible values to find the maximum or minimum value.   In workflows available on Biosecurity Commons, you can run a surveillance design that either minimises costs, maximises benefits, or maximises detection. The allowed subset of data is defined by the constraints set by you the user.


The example spatial surveillance workflow which is currently available on Biosecurity Commons is based on a recent paper exploring how to allocate surveillance efforts for Orange Hawkweed in a relatively small area in Victoria (Hauser and McCarthy 2009). This workflow was also based on these papers (Anderson et al. 2017, McCarthy et al. 2010, and Moore et al. 2016)


This methodology is about finding the best way to survey and manage invasive species in an area. The area is divided into equal-sized sites (each grid cell is a site), and each site has a certain probability of having the invasive species. To estimate this spatial variation in probability, factors such as habitat, past invasive species presence, and distance from known invasions can be considered. This probability could come from an SDM model, a probability from a risk map, or as in the case in this example the probability based on results from a dispersal model. 

Detecting an invasive species depends on the effort put into the search, the effectiveness of the detection method, and the site's terrain. In our example detection probabilities vary spatially, but costs could also vary spatially. The probability of failing to detect an invader decreases as search effort increases. 

When an invader is detected, management efforts focus on eradicating it, which can include destroying or removing the species and monitoring the area to make sure it's gone. These efforts come with costs, like staffing, equipment, and potential damage to the environment or agriculture. 

If the species isn't detected during the current survey, it could spread and cause more damage, making it harder to eradicate later. This study aims to find the best way to balance the costs of surveillance and management while minimizing the total expected costs. 

To do this, the researchers use mathematical equations that consider factors like the probability of the invader being present, the effectiveness of surveillance methods, and the costs associated with managing the invasion. They find the best balance between surveillance and management costs by optimizing these equations. 

Sometimes, there's a limited budget for surveillance, so the researchers also came up with a method to distribute the budget among the different sites based on factors like the probability of invasive species presence and the effectiveness of surveillance methods. This method involves ranking sites by priority and calculating the best allocation of resources based on the budget and other factors. 

In summary, this methodology helps find the best way to spatially invest resources in surveillance and management efforts to minimize the total costs of dealing with invasive species, while also considering budget constraints. It does so by dividing an area into smaller sites, estimating the probability of invasive species presence, and optimizing equations that balance surveillance and management costs. 

We envision the surveillance workflow being used for two primary purposes, either early detection or delimitation.  



Early Detection


Optimal surveillance designs often seek to identify incursions early so they are easier and cheaper to control or eradicate. The optimization solutions look to balance the costs associated with surveillance against the costs associated with undetected incursions that go on to become widely distributed (Hester et al. 2017, Quinlan et al. 2015). 


Table 1. Literature on approaches to surveillance design for early detection.





Optimal surveillance intensity for minimising total cost. 

Simulates invasion, growth and detection to determine optimal surveillance intensity that minimises total management costs across multiple regions. 

 Markov simulations; 

budget constrained; uses solver. 


Bogich et al. 2008;  Epanchin-Niell et al. 2012; Epanchin-Niell et al. 2014

“Search and destroy”: cost-effective surveillance allocation. 

Method to determine surveillance and treatment allocation that minimises management costs. 

Site prioritisation; optimal allocation; simple algorithm; 

budget constrained. 


Hauser & McCarthy 2009

Optimal surveillance allocation via zonation. 

Method uses the zonation algorithm to prioritise sites (cells) by iteratively removing sites with minimal benefit. 

Site prioritisation; optimal allocation;  multispecies; prioritisation function alternatives; incorporates costs. 

Camac et al. 2021; Moilanen et al. 2005

Optimal allocation of multiple surveillance components over risk zones. 

Method for allocating multiple surveillance components to risk zones based on cost to provide a minimum design prevalence (Barrow Island case study). 

Multiple surveillance system components (SSC); risk zones; 

design prevalence; 

probabilistic approach. 

Barrett et al. 2010; Jarrad et al. 2011; Murray et al. 2015; Whittle et al. 2013

Detection survey design. 

Provides methods for calculating the sample sizes required to achieve a specified level of confidence in detection. 

One and two-stage sampling; design density; design prevalence. 

Kean et al. 2015

Surveillance design assessment via hierarchical Bayesian modelling. 

Bayesian-based stochastic simulations of pest invasions and detection for assessing surveillance design scenarios. 

Markov chain Monte Carlo (MCMC) simulations; 

observation model; invasion process model. 

Stanaway et al. 2011; Stanaway 2015

Optimal allocation of fixed surveillance resources. 

Computational algorithms for optimising surveillance via various criteria. 

Optimises detection time, detection probability, or surveillance benefit; iterative algorithms; resource constraints. 

Cannon 2009

Simulations for optimal surveillance efforts. 

Spatially-explicit simulations utilised to optimise surveillance strategies and parameters. 

Models passive and active surveillance; spread simulations; surveillance scenarios; cost constrained; optimisation via genetic algorithm. 

Cacho et al. 2010; Hester and Chacho 2012




Optimal surveillance designs can also seek to identify the full extent of an incursion. The delimited extent of incursions is required to inform containment and elimination strategies. Subsequent delimitation informs ongoing surveillance for monitoring the progress of pest management activities, as well as providing evidence for reinstating area freedom for trade purposes (Hester et al. 2017, Quinlan et al. 2015). 


Table 2. Literature on approaches to surveillance design for delimitation.






Delimitation evaluation method. 

Method for delimiting the extent of an incursion to evaluate eradication or containment progress and success. 

Delimitation (D) measures; extirpation  (E) measure; 

Eradograph (D vs E); trends indicative progress; informs responses. 

Panetta and Lawes 2005; Panetta and Lawes 2007;  Burgman et al. 2013

Surveillance prioritisation for delimitation.  

Method for optimal prioritisation of delimitation survey effort (extends [9]). 

Site prioritisation targets areas outside incursion; optimal allocation; budget constrained. 

Hauser et al. 2016  

Approach, decline, delimit (ADD). 

Utilises the rate of decline in occurrence to delimit the likely invasion extent. 

Samples along transects; decline in sampling (outward); estimates prevalence; decline curve fitting. 

Leung et al. 2010





Example of how to produce an optimal spatial surveillance 


Below we highlight how to run a spatial surveillance problem like the one presented in a paper on Orange Hawkweed in a relatively small area in Victoria (Hauser and McCarthy 2009).  Again, please get in touch with us if you have a different kind of surveillance problem you would like to optimize so we can tailor our platform to address your question. 


Step 1 – Review available evidence, knowledge, and data


Keep in mind it is easy to get a result by using the dashboard tools available at Biosecurity Commons, but any optimal design of where to deploy surveillance efforts will be greatly improved by focusing efforts on generating the best available input data. In spatial surveillance that includes having a robust estimation of both the likelihood of pest establishment and the efficacy or detection rates at each site (or grid cell). While we are developing tools to assist with constructing these kinds of data, the quality of results from this workflow will be fully dependent on the quality of these inputs.  Other things to consider if working in grids would be to select a cell size that best reflects variation in how surveillance activities would be allocated, to estimate costs at each site (or grid cell) and consider the objective function and constraints carefully. Ultimately, the utility of a surveillance optimization problem will be fully dependent on the data used and the way in which you set up the problem.


Step 2 –  Provide the context of your surveillance questions


Taking time to fill out the context allows users to capture the kind of surveillance workflow that was run as well as the original units used within the experiment’s data.  This can be helpful to record if a user is returning to a saved project or for users sharing a project. These setting do not alter the results. 

Surveillance type: “survey” Other types of surveillance that could be used in clude traps, samples, reports, mixed and other. 

Surveillance quantity unit: “hours” Other types of units which could be used to describe the quantity of surveillance to be allocated to each site (or grid cell) include units, traps, detectors, or samples. 

Cost unit: “hours”  The descriptive unit to describe costs or benefit savings can be described as either hours or dollars “$”. 

Distance/area unit: “meters” The unit to describe spatial distance can be set to either meters or kilometers. 

Time unit: “years” The units used to describe time could also include: months, weeks, days or hours. 

Step 3 –  Select the method of surveillance design 

Currently, two types of methods are available, “spatial surveillance” or “sampling surveillance” and in this example we select spatial surveillance where inputs and rasters are spatially explicit rasters. Sampling surveillance simply uses CSV files to identify constraints and results. Both methods use Lagrange-based optimization methods. 

In this example: 

Method: “Spatial Surveillance” 

Step 4 –  Select the input parameters for the optimisation problem 

Select the rasters with spatial variation in occurrence and detection probabilities, the objective function and constraints (Anderson et al. 2017, Hauser and McCarthy 2009, McCarthy et al. 2010, and Moore et al. 2016). 

Region: “Bogong High Plains template WGS84” Select the raster that has the extent, resolution and coordinate reference system to be used in for all rasters. Other rasters will be transformed to match the raster selected here.  

Users can also select a “Custom Region” by selecting an existing pre-defined boundary such as a NRM or state boundary. Or can draw on a map, and then select a raster with the desired resolution and CRS with an extent at least as large as the area selected as a custom region. 

Occurrence probability: “Orange hawkweed occurrence probability WGS84” Select the raster with values that represent the likelihood of pest occurrence at each site “grid cell”. This layer could be taken from results from a species distribution model, a risk map, or a dispersal model. 

Users can also select a function to generate an occurrence raster. 

Efficacy (lambda): “Orange hawkweed surveillance efficacy WGS84” Select the raster which indicates the variable expected rate of detection. Here the expected detection rate is simply low in shrubby areas and relatively high in areas where the vegetation is low and grassy.  The number in each site (grid cell) is the expected number of detections per “Surveillance quantity unit” in this case per hour. 

Users can also select a “Transform Layer” function to perform a (linear, exponential, logarithmic, upper or lower) transformation on an existing raster layer. 

Optimisation strategy: “cost” This is the objective function and in this case the objective is to minimise cost.  Other options include to maximise the benefit or to maximise the detection. 

Management cost: 

Detected: “1000” The management cost (in hours – the cost unit) of finding the weed. 

Undetected: “10000” A ten-fold increase in the management cost (in hours- the cost unit) of failing to detect the weed when it was present. 

Allocated surveillance cost: N/A – The option to add a layer that specifies the per unit of cost of conducting surveys within the site. In this example it would be the cost per hour of conducting surveys in each grid cell (site). 

Fixed cost: N/A – The option to add a layer that specifies a fixed per unit cost associated with surveying at that site. In this example it could be the cost in hours of travelling to each grid cell (site). 

Constraint: “budget” Users could also select “confidence” i.e. 0.95 or “none”. 

Surveillance Budget: “1125” total number of hours (cost unit) allowed for surveillance. 

Minimum Allocation: “N/A” Left blank here, but the minimum number of hours that would be considered a minimum practical allocation to warrant visiting the site.  For example, an allocation of less than 1 minute might be impractical. 

“unticked” Discrete Allocation If ticked the allocation units are set to discrete integers, in this case, if it had been ticked each value would be 1, 2, 3, 4, 5 etc hours. It would be appropriate to use this if each the allocation units were traps, or detectors. 



Literature cited: 


Anderson, D. P., Gormley, A. M., Ramsey, D. S. L., Nugent, G., Martin, P. A. J., Bosson, M.,Livingstone, P., & Byrom, A. E. (2017). Bio-economic optimisation of surveillance to confirm broadscale eradications of invasive pests and diseases. Biological Invasions, 19(10), 2869–2884. doi:10.1007/s10530-017-1490-5

Barrett, S., Whittle, P., Mengersen, K., & Stoklosa, R. (2010). Biosecurity threats: the design of surveillance systems, based on power and risk. Environmental and Ecological Statistics, 17(4), 503–519. https://doi.org/10.1007/s10651-009-0113-4

Burgman, M. A., McCarthy, M. A., Robinson, A., Hester, S. M., McBride, M. F., Elith, J., & Panetta, F. D. (2013). Improving decisions for invasive species management: reformulation and extensions of the Panetta—Lawes eradication graph. Diversity and Distributions, 19(5/6), 603–607. https://www.jstor.org/stable/23479781 

Cacho, O. J., Spring, D., Hester, S., & Mac Nally, R. (2010). Allocating surveillance effort in the management of invasive species: A spatially-explicit model. Environmental Modelling & Software, 25(4), 444–454. https://doi.org/10.1016/j.envsoft.2009.10.014 


Camac, J., Baumgartner, J., Hester, S., Subasinghe, R., & Collins, S. (2021). Using edmaps & Zonation to inform multi-pest early-detection surveillance designs. Tech. Rep. 20121001, Centre of Excellence for Biosecurity Risk Analysis. https://cebra.unimelb.edu.au/__data/assets/pdf_file/0009/3889773/20121001_final_report.pdf


Cannon, R. M. (2009). Inspecting and monitoring on a restricted budget-where best to look? PREVENTIVE VETERINARY MEDICINE, 92(1–2), 163–174. https://doi.org/10.1016/j.prevetmed.2009.06.009 


Epanchin-Niell, R. S., Haight, R. G., Berec, L., Kean, J. M., & Liebhold, A. M. (2012). Optimal surveillance and eradication of invasive species in heterogeneous landscapes. ECOLOGY LETTERS, 15(8), 803–812.   https://doi.org/10.1111/j.1461-0248.2012.01800.x 


Epanchin-Niell, R. S., Brockerhoff, E. G., Kean, J. M., & Turner, J. A. (2014). Designing cost-efficient surveillance for early detection and control of multiple biological invaders. Ecological Applications, 24(6), 1258–1274. https://doi.org/10.1890/13-1331.1 

Hauser, C. E., & McCarthy, M. A. (2009). Streamlining ’search and destroy’: cost-effective surveillance for invasive species management. Ecology Letters, 12(7), 683–692. doi:10.1111/j.1461-0248.2009.01323.x


Hauser, C. E., Giljohann, K. M., Rigby, M., Herbert, K., Curran, I., Pascoe, C., Williams, N. S. G., Cousens, R. D., & Moore, J. L. (2016). Practicable methods for delimiting a plant invasion. Diversity and Distributions, 22(1/2), 136–147. https://doi.org/10.1111/ddi.12388 


Hester, S., & Cacho, O. (2012, January 1). Optimization of Search Strategies in Managing Biological Invasions: A Simulation Approach. HUMAN AND ECOLOGICAL RISK ASSESSMENT, 18(1), 181–199. https://doi.org/10.1080/10807039.2012.632307

Hester, S., Hauser, C., & Kean, J. (2017). Tools for Designing and Evaluating Post-Border Surveillance Systems. In A. Robinson, T. Walshe, M. Burgman, & M. Nunn (Eds.), Invasive Species: Risk Assessment and Management (pp. 17-52). Cambridge: Cambridge University Press. https://doi.org/10.1017/9781139019606.003 


Jarrad, F. C., Barrett, S., Murray, J., Parkes, J., Stoklosa, R., Mengersen, K., & Whittle, P. (2011). Improved design method for biosecurity surveillance and early detection of non-indigenous rats. New Zealand Journal of Ecology, 35(2), 132–144. https://www.jstor.org/stable/24060661

Kean, J. M., Burnip, G. M. & Pathan, A. (2015). Detection survey design for decision making during biosecurity incursions. In F. Jarrad, S. Low- Choy & K. Mengersen (eds.), Biosecurity surveillance: Quantitative approaches (pp. 238– 250). Wallingford, UK: CABI. https://doi.org/10.1079/9781780643595.0000 


Leung, B., Cacho, O. J., & Spring, D. (2010). Searching for non-indigenous species: rapidly delimiting the invasion boundary. Diversity and Distributions, 16(3), 451–460. https://doi.org/10.1111/j.1472-4642.2010.00653.x 

McCarthy, M. A., Thompson, C. J., Hauser, C., Burgman, M. A., Possingham, H. P., Moir, M. L.,Tiensin, T., & Gilbert, M. (2010). Resource allocation for efficient environmental management. Ecology Letters, 13(10), 1280–1289. doi:10.1111/j.1461-0248.2010.01522.x

Moilanen, A., Franco, A. M. A., Early, R. I., Fox, R., Wintle, B., & Thomas, C. D. (2005). Prioritizing Multiple-Use Landscapes for Conservation: Methods for Large Multi-Species Planning Problems. Proceedings: Biological Sciences, 272(1575), 1885–1891. https://doi.org/10.1098/rspb.2005.3164 

Moore, A. L., McCarthy, M. A., & Lecomte, N. (2016). Optimizing ecological survey effort overspace and time. Methods in Ecology and Evolution, 7(8), 891–899. doi:10.1111/2041-210X.12564


Murray, J., Whittle, P., Jarrad, F., Barrett, S., Stoklosa, R., and Mengersen, K. (2015). Design of a Surveillance System for Non-indigenous Species on Barrow Island: Plants Case Study. In F. Jarrad, S. Low- Choy & K. Mengersen (eds.), Biosecurity surveillance: Quantitative approaches (pp. 203–216). Wallingford, UK: CABI. https://doi.org/10.1079/9781780643595.0000 


Panetta, F. D., & Lawes, R. (2005). Evaluation of Weed Eradication Programs: The Delimitation of Extent. Diversity and Distributions, 11(5), 435–442. https://www.jstor.org/stable/3696928 


Panetta, F. D., & Lawes, R. (2007). Evaluation of the Australian Branched Broomrape (Orobanche ramosa) Eradication Program. Weed Science, 55(6), 644–651. https://www.jstor.org/stable/4539630 


Quinlan, M., Stanaway, M., and Mengersen, K. (2015). Biosecurity Surveillance in Agriculture and Environment: a Review. In F. Jarrad, S. Low- Choy & K. Mengersen (eds.), Biosecurity surveillance: Quantitative approaches (pp. 9–42). Wallingford, UK: CABI. https://doi.org/10.1079/9781780643595.0000

Stanaway, M. A., Mengersen, K. L., & Reeves, R. (2011). Hierarchical Bayesian modelling of early detection surveillance for plant pest invasions. Environmental and Ecological Statistics, 18(3), 569–591. https://doi.org/10.1007/s10651-010-0152-x  

Stanaway, M. (2015). Evidence of Absence for Invasive Species: Roles for Hierarchical Bayesian Approaches in Regulation. In F. Jarrad, S. Low- Choy & K. Mengersen (eds.), Biosecurity surveillance: Quantitative approaches (pp. 265–277). Wallingford, UK: CABI. https://doi.org/10.1079/9781780643595.0000 

Whittle, P. J. L., Stoklosa, R., Barrett, S., Jarrad, F. C., Majer, J. D., Martin, P. A. J., & Mengersen, K. (2013). A method for designing complex biosecurity surveillance systems: detecting non-indigenous species of invertebrates on Barrow Island. Diversity and Distributions, 19(5/6), 629–639. https://www.jstor.org/stable/23479784