The German wasp, Vespula germanica (Fabricius, 1793) (Hymenoptera, Vespidae), is an agricultural, environmental, and social pest that has been introduced to the Southern Hemisphere on various occasions over the last century (Lester and Beggs 2019; Wilson Rankin 2021). The increased movement of people and cargo after World War II saw V. germanica spread from its native range in Europe and the Mediterranean region to North America, Argentina, Chile, Iceland, South Africa, New Zealand, and Australia, where it has become invasive (Gergonne et al. 2026). It was first reported in the Western Cape region of South Africa in 1972, although it could have been present before 1970 (Whitehead 1975; Whitehead and Prins 1975). It is currently listed as a category 1B invasive species in South Africa, which necessitates a management plan, but eradication is probably feasible given the relatively slow spread of the species until recently (van Zyl et al. 2018; Veldtman et al. 2021). This slow spread rate is unusual, since all reports of V. germanica in South Africa are surrounded by climatically optimal and highly suitable habitat (de Villiers et al. 2017).

It is not known how V. germanica first arrived in the Western Cape region, but because only two haplotypes have been identified, denoted Hap1 and Hap2, it may be that two incursion events led to successful establishment (van Asch et al. 2022; Gergonne et al. 2026). Hap2 was most likely involved in the first incursion, since it is the dominant haplotype around the original invaded range in and around the Cape Peninsula, whereas Hap1 appears to have arrived in the Boland region northeast of Cape Town in 2000 and subsequently spread to the north, east, and south (Davies et al. 2020; Veldtman et al. 2021) (Fig. 1). It is therefore hypothesized that Hap1 is more climatically suited to the Western Cape region than Hap2 (van Asch et al. 2022).

Given its accelerated spread in the past 20 years, V. germanica could cause substantial economic costs to apiculture and fruit and vegetable industries. Bees from both wild and managed European honeybee (Apis mellifera Linnaeus) colonies are attacked by the wasp, sometimes by direct raids on beehives but more frequently by attacks on individual adult bees at ground level (Pusceddu et al. 2017). The resultant decline in free and commercial pollination services is particularly serious in the case of South Africa, where there are currently insufficient managed hive rental services available to fruit and vegetable growers (Allsopp et al. 2008; Veldtman 2018), and only enough wild pollinators to satisfy 20% of horticultural pollination needs (Allsopp 2020). It is therefore reasonable to assume that the demand for pollination services is highly inelastic. Any reduction in commercial hive capacity caused by V. germanica will produce a proportional increase in the price of hive rentals for fruit and vegetable growers, increasing their variable costs of production.

In this paper, a model is described that was used to predict the likely economic cost of a spreading V. germanica population in South Africa. If the population has entered an accelerated growth phase, it is important to highlight the potential costs that could occur as a result, thereby communicating to policymakers the benefits of taking management actions now before the scale of the outbreak becomes unmanageable. This can happen rapidly, as overwintering nests can reach large sizes by the following summer and produce thousands of new queens (Tennant et al. 2011), and as new colonies take hold in an area, their foraging behavior will begin to affect the surrounding environment. Pollination services to horticulture will be affected as V. germanica attacks managed and wild honeybee hives; potential feeding by wasps on fruit will affect wine grape growers; and household costs will rise due to the need to remove nests and avoid disruption to recreational activities. These effects were simulated, and their financial costs were summed over 50 years to highlight the economic significance of this pest and the potential return on investment in actions that slow or stop its spread. All costs are stated in United States dollars (US$).

The extent of losses experienced in each sector is expressed in appropriate units in Fig. 2, which includes the effects of V. germanica on commercial beehives. Financial losses to the pollination industry were not quantified in the assessment but were assumed to affect horticulture producers indirectly, as they were passed on as increased hive rental costs and yield losses from wasps attacking bees in beekeepers’ apiaries. The number of commercial beehives affected over time is therefore shown in panel A, while the area of horticulture affected is shown in panel B. The area of wine grapes affected is shown in panel C, and the number of households likely to be affected is shown in panel D. The box-whisker plots show the 5^th, 25^th, median, 75^th, and 95^th percentile values.

With the set of assumptions used in the model, it was estimated that 6,500–20,400 beehives would be affected by V. germanica attacking foraging bees after 50 years (panel A). This range corresponds to the 25^th and 75^th percentiles, indicating that 50% of model iterations produced results between these values. Likewise, the results showed that 127,000–136,000 ha of horticulture crops would be affected by yield losses and rising production costs (panel B), 10,000–19,000 ha of wine grapes would be affected by higher production costs because of wasp control expenses (panel C), and 109,000–120,000 households would experience increased pest removal expenses (panel D) by the end of the simulation period.

The predicted indirect effects on horticulture from induced changes in the commercial pollination industry are shown in Fig. 3. Panel A shows the change in the price per hive paid for pollination services in real or present-value terms. As foraging bees are attacked by spreading V. germanica colonies, the resulting contraction in supply places upward pressure on the hire price. This effect had a negligible impact on hive hire prices for the first 25–30 years but increased rapidly thereafter. After 50 years, the estimated price per hive increased from US$75.00 to US$78.20–84.50. Panel B shows the estimated yield lost in horticulture industries due to V. germanica attacking bees. This had the same lag before losses accelerated, with yield losses increasing by 4–13% by the end of the 50-year simulation.

The present value of damages predicted in the model simulations (i.e., PVd_T) is shown in Fig. 4. Panels A, B, and C show predicted V. germanica impacts on growers of pollinator-dependent horticulture crops, wine grape growers, and householders, respectively, while panel D shows the summed value of damages across all three sectors.

Disruption to honeybee pollination services is likely to cost horticulture industries US$30–55 million over 50 years. Costs related to wine grape production over the same period were estimated to be US$1–3 million, while costs incurred by households were estimated to be US$110–260 million. The summed value of damages from V. germanica across all sectors was US$160–300 million after 50 years, with a median value of US$225 million (Table 3).

Given the uncertainties in many model parameters, a sensitivity analysis was conducted to determine which exerted the greatest influence on the results. A simple approach was used, involving each parameter being set to the minimum and maximum values of its range while all other parameters were held constant. The effect that the minimum and maximum values for each parameter had on the present value of combined costs is reported in Fig. 5. Here, the influence of each parameter is indicated by the length of the corresponding bars, and the parameters are ranked from top to bottom in order of the change they produced in the present value of cumulative costs after 50 years.

The results were most sensitive to changes in the rate at which V. germanica was assumed to spread. The estimation technique used to shape the logistic spread function (Eq. 2) specified a proportion of each sector, denoted θ, close to the asymptote to indicate where the function began to plateau. The number of time steps taken before V. germanica affected θ, t^θ, determined the rate of spread. For all sectors, t^θ was assumed to be 30–50 years (Table 1). Increasing t^θ to 50 years lowered the median value of damages from US$225 million to US$115 million (i.e., –50%), and lowering t^θ to 30 years increased the median value of damages to US$345 million (+55%).

Large changes in the results were also observed when varying the discount rate. The simulation period was relatively long at 50 years, and cost accumulation was skewed to later time steps, so the erosive effects of discounting were pronounced. Lowering the discount rate to 5% from its median value of 6% (i.e., –17%) produced an increase in the median value of damages to US$335 million (+49%). Likewise, increasing the discount rate to 7% (+17%) lowered the present value of costs to US$160 million (−28%). The discount rates used were suggested in the Commonwealth of Australia (2006), but it was acknowledged that choosing an appropriate value for this parameter in a case study with cost streams that extend so far into the future is difficult (Abelson and Dalton 2018).

Results were also relatively sensitive to changes in the parameter θ, which was used to determine the asymptotic phase of the spread function. This was assumed to be 50–60% for horticulture, 10–20% for wine grapes, and 70–80% for households. Increasing θ to the maximum of its range for each sector increased the median value of damages to US$280 million (i.e., +25%), and lowering θ to the minimum value of its range for each sector lowered the median value of damages to US$190 million (−15%).

The model outlined in this paper is a first attempt to estimate the potential financial impacts V. germanica could have in South Africa if it continues to spread unabated over the next 50 years. Although relatively simple, it captures important spread characteristics that make this a difficult species to deal with, namely its propensity to establish and maintain a relatively low economic cost for several decades before population growth and damage accelerate. The model aggregates economic impacts over time, indicating the financial implications of the spread to communities and agricultural industries. To this end, it provides information that could be used in future decisions concerning national response policies.

There is a growing literature highlighting increasing economic costs associated with invasive species, but this often relies on aggregated impacts from multiple species (Early et al. 2016; Paini et al. 2016; Cuthbert et al. 2022; Polce et al. 2023; Gallardo et al. 2024; Soto et al. 2025). These are frequently cited in the introductory sections of many research papers and articles to highlight the scale of the issue, but beyond this, their use is questionable. Policymakers are generally aware that the invasive species problem is dire and the costs are enormous, but they are seldom informed by predictive analyses of what the effects of response policies enacted in the present might be (Regan et al. 2002; Waage et al. 2005).

The model provides policymakers with prospective policy analysis. Although this can be achieved using simple partial budgeting techniques or deterministic models (de Wit et al. 2001; MacLeod et al. 2004; Bolda et al. 2010), more complex approaches such as the one set out in this paper rely on stochastic simulation to specifically incorporate policy risks (Rafoss 2003; Yemshanov et al. 2009; Cook 2019). Other approaches that integrate ecological models with economic management frameworks (Alexei and Andrew 1998; Sharov and Liebhold 1998; Barbier 2001; Cacho et al. 2008; Hyder et al. 2008; Carrasco et al. 2010 ) or use metapopulation models to forecast future impacts (Albers et al. 2010; Sanchirico et al. 2010; Pepper et al. 2019) can also highlight complex interactions between spreading populations and heterogeneous landscapes to policy audiences. However, these models can be expensive to develop and require technical skills to use, and their complexity can make results difficult to communicate to policymakers.

In the current paper, it was assumed that no coordinated eradication or spread-mitigation response will be mounted as V. germanica spreads in South Africa over the next 50 years, but the model can be used to evaluate future response policies. For instance, the potential human health consequences of this pest as it spreads could create a political imperative for authorities to respond, particularly around the 20-year mark when the simulations suggest spread will reach an inflection point. A benefit-cost analysis of any prospective response policy could simply be carried out using these results as a counterfactual scenario and comparing them with the potential impacts of V. germanica in a policy scenario.

A simple example drawing on the results of the sensitivity analysis can be used to explain how a benefit-cost analysis could be carried out. Sensitivity test results (Fig. 5) suggest that if a spread-mitigation policy can increase the parameter tθ to 50 years, the value of damages will be reduced by US$110 million over 30 years, or approximately US$3.7 million per year. This represents the threshold or break-even investment cost for the policy, which might take the form of surveillance activities to detect and destroy wasp nests (Duthie and Lester 2013; Cook 2019), subsidies for pest control services (Saga et al. 2025), bounties for nest reporting (Hester and Cacho 2012), or potentially biological control programs (Rose et al. 1999; Merino et al. 2013; Cacho and Hester 2022; Hester et al. 2024; van Zyl et al. 2024). If the policy costs less than this amount, it will produce a net benefit for society over time, but if not, a net social welfare loss is likely to result. A benefit-cost analysis of this nature would be an important tool for policymakers, as it condenses a great deal of complex information into monetary terms, making it relatively simple to communicate to policymakers and stakeholders (Waage et al. 2005). However, this must be accompanied by a technical feasibility study that assesses logistical constraints involved in applying sustained pressure to wasp populations given their high mobility, high reproductive rate, and ability to exploit a range of food resources (van Zyl et al. 2018).

The simplifying assumption was used in the model that the supply of managed hives remains constant over time despite a rising demand for pollination services. This implies that no new commercial beekeepers will enter the market, even as hive hire prices rise, to increase supply and place downward pressure on prices. As a result, the model predicts the average cost per honeybee hive supplied for pollination to rise from approximately US$75.00/hive to US$78.20–84.50/hive over the next several decades if V. germanica continues to spread (Fig. 3). The assumptions reflect the barriers new suppliers face when entering the market. South Africa has banned the importation of live honeybees since the early 1980s (Dietemann et al. 2009), and the availability of wild bee stock to build or replenish commercial stock is limited (Veldtman 2018; Allsopp 2020). Access to year-round forage resources in South Africa’s dry climate is also limited and subject to seasonal variations that could affect the ability of the commercial pollination industry to expand (Veldtman 2018). Concerted conservation efforts specifically oriented toward plant species that benefit honeybees are needed to provide sufficient forage for wild and managed colonies (Mensah et al. 2017).

If the set of assumptions used to characterize V. germanica spread is correct, or even a reasonable approximation, impacted sectors are likely to begin experiencing substantial costs after 20 years due to the logistic growth assumed in the model. This implies a significant lag period given that the pest has been present in the Western Cape province for many decades. Vespula germanica has taken 50 years to spread from the southern slopes of Table Mountain after crossing a major geographical barrier, the Cape Flats (a large, low-lying region to the southeast of Cape Town), through human-mediated spread 25 years ago. However, since then, spread has been somewhat sporadic. While it quickly proliferated in the Boland region, spread appeared to slow after 2005 before increasing again from 2014, again due to human-mediated spread. This suggests that the increase in area occupied might not be occurring as uniformly as the model predicts, at least in the short term, although the total area affected remains relatively small. The complicated spread pattern observed indicates that a spatial time-step analysis would be beneficial, which the model does not currently include. This would be particularly useful when predicting regional and subregional spread and the economic implications for local agricultural industries and communities.

The lack of a geospatial dimension in the model could be addressed in future work using modeling techniques that delineate landscapes into spatial units, such as species dispersal models or cellular automata models. Dispersal models incorporating the processes regulating species survival, reproduction, and movement have been used effectively to explain the spread of invasive species over large spatial and temporal scales (Pitt et al. 2009; Wittmann et al. 2016; Lustig et al. 2017). However, these have yet to be combined with economic models to inform policy and can be problematic when used for prediction (Liu et al. 2020). Cellular automata models have been effective policy support tools for invasive species preparedness and response efforts when combined with measures of economic impact (Cook et al. 2025) but may be unsuited to diverse landscapes where movement probability varies according to exogenous factors such as wind or human-mediated dispersal (Thompson et al. 2021). Given problems with more complex approaches, the aspatial model presented in this analysis is an important first step in informing policymakers of the potential economic threat posed by V. germanica.

An estimate of the damage likely to be caused by V. germanica if it is allowed to spread in South Africa over the next 50 years has been provided in this paper. By applying a spread model to multiple sectors, it was shown how costs could increase at an increasing rate over time. The effects of wasps preying on foraging honeybees could raise variable production costs and yield losses for horticulture industries, and the value of damages was estimated to be US$30–55 million over 50 years. Cost increases to the wine grape industry from wasp control measures to prevent feeding on grapes were shown to be relatively minor, with costs estimated at US$1–3 million over 50 years. The financial impacts on households were shown to be high, with the predicted increase in pest extermination services in urban areas likely to cost US$110–260 million. These results suggest that the combined costs to these three sectors could be US$160–300 million after 50 years, implying annual costs of US$3–6 million. This implies that if a response was mounted by the South African government to mitigate the spread of V. germanica at a cost of up to US$3 million per year, this would most likely produce a net benefit for society. Further research is needed to expand on the simulation model introduced in this paper. More scientific investigation into the spread of V. germanica, particularly Hap1, would be extremely helpful to improve the model, given that results were highly sensitive to parameters that affected the spread rate. Introducing a spatial dimension to explore interactions between different habitats and spreading wasp populations may help in the design of future responses at a regional and subregional scale.