Spread of invasive ragweed: climate change, management and how to reduce allergy costs

Authors


Summary

  1. Ragweed Ambrosia artemisiifolia L. is rapidly spreading in Europe. Its pollen is highly allergenic, with 4–5% of Europeans being sensitized. There is an urgent need to curtail the further spread to minimize allergy costs.
  2. We simulated the spread of ragweed in Austria and southern Germany (Bavaria) until 2050 with particular emphasis on expected climate change. Using different management scenarios and levels of management effort, we analysed the potential for reducing human allergy costs, that is, expenses caused by allergies from ragweed pollen, by curtailing the accelerating spread of ragweed. We accounted for three contrasting climate assumptions: no change in temperature and moderate (annual temperature increase of 0·025 °C) and more extreme (annual temperature increase of 0·04 °C) climate change.
  3. We found that a carefully designed management plan consisting of survey and eradication can drastically reduce the spread of ragweed. Without management, mean allergy costs for the management period (2011–2050) amount to about 290, 335 and 365 million € annually under the three climate change assumptions.
  4. Following an optimally allocated management strategy with an annual budget of 30 million € reduces mean allergy costs by 258, 295 and 325 million € per year. Thus, the management may yield substantial savings, in particular under more extreme warming, where total savings over 40 years amount to about 12 billion €.
  5. Synthesis and applications. Our study illustrates that management of invasive alien species has an economic benefit beside natural conservation. We provide guidance for the future management using the example of ragweed in Austria and Bavaria and show that although the species has expanded its range and abundance substantially in recent years, a well-designed and ambitious management programme still may yield substantial benefits. This is true for current climatic conditions as well as for future climate change scenarios, albeit management costs increase with a warming climate. However, possible gains are increasing in parallel. Given the scale of impacts on human health, and the substantial gains accrued from management, our results suggest that it is wise to halt further spread of ragweed.

Introduction

Species distributions are codetermined by climatic factors. Expected climate change will therefore have a major impact on species abundance and ranges (Thuiller et al. 2005; Petitpierre et al. 2012) irrespective of whether species are native or introduced to a particular region. In Europe, many alien plants and animals are assumed to be geographically restricted by current climatic conditions (Walther et al. 2009). Climate warming may therefore trigger or accelerate the spread of such species in hitherto unaffected regions or foster population growth in already occupied areas and, consequently, reinforce their negative impacts on biodiversity and human well-being (Vilà et al. 2010, 2011). Among these alien species, ragweed Ambrosia artemisiifolia L. is considered particularly relevant in terms of impact on human health and agriculture, both in Europe (Brandes & Nitzsche 2007; Essl, Dullinger & Kleinbauer 2009; Vilà et al. 2010; Bullock et al. 2012) and in temperate North America (Ziska et al. 2011). Ragweed is a highly allergenic, wind-pollinated herb with an annual life cycle. It is native to parts of North America and was first introduced to Europe during the 19th century (Brandes & Nitzsche 2007). After an extended lag phase, ragweed has been spreading since the mid-20th century and is now a pest in several countries of Eastern Europe and rapidly expanding in Central Europe (Chauvel et al. 2006; Brandes & Nitzsche 2007; Dullinger et al. 2009; Smolik et al. 2010; Richter et al. 2012). A single ragweed specimen can produce up to 50 000 seeds (Brandes & Nitzsche 2007) and on average about 1 billion pollen grains (Fumanal, Chauvel & Bretagnolle 2007). These numbers might even rise in the future as climatic warming has been shown to substantially increase ragweed pollen production and resulting pollen loads (Ziska et al. 2011).

Implementing any effective survey and eradication (‘management’) programme for ragweed has to address a range of complementary fields of action. Few regulations of these fields are currently in place, for example, a recent amendment of the European Union on improved seed regulation (EU 2011) has introduced strict rules for bird seed imports, which formerly often were contaminated by ragweed seeds (Vitalos & Karrer 2008). Mowing of infested embankments of major roads has been adjusted in parts of Austria to prevent ragweed plants from completing their life cycle and set seeds (Karrer et al. 2011), and several cities and regional authorities have started information campaigns and encourage the weeding of infestations by landowners (StMUG 2012). However, other elements of a comprehensive management strategy including a coordinative institution of activities on the national level are currently not in place, limiting the efficacy of current (and future) management actions.

Recently, Smolik et al. (2010) presented a spatially and temporally explicit framework that allows the spread of ragweed to be simulated in discrete annual steps across a gridded landscape. Based on that work, Richter et al. (2012) reconstructed range dynamics of ragweed in Austria and subsequently identified optimal management strategies to limit the species’ further spread across Austria and southern Germany (Bavaria). These authors considered budget constraints and asked whether there is an optimal minimum habitat suitability threshold for distinguishing the areas to be covered by surveillance and eradication measures from those that do not warrant search and control efforts, that is, a minimum suitability that maximizes management success under given budget constraints. In addition, they studied whether site suitability may also provide a guideline for the sequence of management activities when budgets do not allow for surveying all the area susceptible to invasion of the species each year. Following a sequence of suitability ranks (for definition see Richter et al. 2012) when targeting management turned out to be more useful than selecting the areas at random from a given set of grid cells. However, these simulations of ragweed spread did not calculate concrete cost-benefit ratios of management and were conservative as they disregarded climate warming.

Here, we relax the assumption of a constant climate and assess, first, the possible consequences of climate change on the spread of ragweed in Austria and Bavaria. Subsequently, we combine these spread projections with pollen dispersion and human population data for estimating consequential allergy costs arising from allergies. Finally, we search for the optimal management strategy that leads to the largest economic reduction on expected health impacts.

Materials and methods

Distribution Data

We extracted all available records of ragweed in Austria and adjacent southern Germany (Free state of Bavaria) until 2005 from the data bases of the project Floristic Mapping of Central Europe (FMCE) in Austria (Niklfeld 1998) and Bavaria (Schönfelder 1999). This project systematically collects distribution data of all vascular plant species on a regular raster of 3 × 5 geographical minutes (~35 km²). In addition to the FMCE data, we searched herbaria as well as the floristic literature (see Essl, Dullinger & Kleinbauer 2009 and Richter et al. 2012 for details). The locality of each additional record was assigned to a grid cell of the FMCE. The date (=year) of each record was taken from the FMCE data base or by consulting the original source or the responsible botanist. We note that this data set does not accurately represent the spatio-temporal invasion process because not all grid cells have been surveyed each year. Also, as our cells have a comparatively large size, we cannot claim that cells without observations actually are completely free of A. artemisiifolia. Our data should therefore be interpreted as distinguishing cells with sizable populations (henceforth ‘infested’) from cells with no or negligible populations (‘uninfested’).

To account for ragweed immigration from outside the study area in our simulation model (see below), we additionally compiled distribution data for those adjacent regions, for which floristic mapping schemes or distribution data bases were available (these are Hungary, Slovenia and South Tyrol). These data have also been assigned to a grid cell of the FMCE.

Spread Model

In the previous work (Richter et al. 2012), we presented a spread model based on the known distribution of ragweed in Bavaria and Austria by 2005. We described the habitat suitability H(x) by a simple generalized linear model (GLM) represented as a logistic function of environmental variables on a grid, namely mean annual temperature νT, annual precipitation sums νP, summed length of major streets νS and land use, that is, the proportion of urban areas and agricultural fields, νL

display math(eqn 1)

with the linear predictor

display math(eqn 2)

We described the spread of ragweed by a leptokurtic dispersal kernel typical for invading organisms (Kot, Lewis & van den Driessche 1996). In our model, the state of each single cell could be occupied or unoccupied and can change at every time step. We did not consider local extinction (i.e. an occupied cell becoming unoccupied) except by eradication. We obtained model parameters by maximum likelihood estimation from the collected spread data (cf. Richter et al. 2012).

Species Spread Promoted by Climate Change

Our present work is based on Richter et al. (2012), but relaxes the assumption of constant climatic conditions. In our simulations, we assume that habitat suitability changes only because of temperature rise, with all other site conditions presumed to remain constant. Different features of individual sites might, however, be spatially correlated. In Central Europe, for example, land use practices and intensity are usually linked to temperature gradients, with highest intensities in regions with highest temperatures. As a consequence, each of the regression coefficients of land use (νL) and temperature (νT), αT and αL, actually accounts for the species’ dependence on both variables to a certain extent (‘shared explained variance’). When modelling a species’ response to changes in one of these variables only, it is hence useful to correct for this hidden dependence on correlated variables. In our simulations, we account for this interdependence by calculating a corrected proportionality factor for temperature increase

display math(eqn 3)

where cov and var denote the covariance and variance of the environmental factors, respectively, so that the difference in h(x) of a given cell is related to the difference in temperature

display math(eqn 4)

In equation (eqn 3), the factor cov(νi, νT)/varT) quantifies the linear variation of νi per variation in temperature. Multiplied by αi this is therefore the variation in h(x) per variation in temperature that is implicitly contained in the dependence on νi. Therefore, the above expression gives the correct proportionality constant for the effect of a variation in a given environmental factor under the condition that all other factors remain constant. On calculating the parameters αi and inline image (see Table 1), we allow for a climate warming trend to start in 1990.

Table 1. Values of the habitat parameters as estimated by maximum likelihood methods from the observed range dynamics of the species in Austria and Bavaria between 1990 and 2005. αi are the weights of the environmental variables used to characterize habitat suitability (equation (eqn 2)); αT, αP, αS and αL correspond to mean annual temperature, annual precipitation, the length of main streets and proportion of urban areas and agricultural fields within the cell, respectively. inline image is the proportionality factor for temperature increase
  α 0 αT(1/°C)αP(1/mm)αS(1/km) α L inline image
Without climate change−10·3 ± 1·270·57 ± 0·100·00338 ± 0·000740·074 ± 0·0141·56 ± 0·43 
ΔT = 0·025 °C per year−10·8 ± 1·200·61 ± 0·080·00313 ± 0·000660·068 ± 0·0161·35 ± 0·400·59 ± 0·09
ΔT = 0·04 °C per year−11·1 ± 1·080·63 ± 0·070·00307 ± 0·000560·065 ± 0·0141·23 ± 0·380·60 ± 0·08

We assume three alternative climatic futures until the mid-21st century: no temperature increase, a moderate increase (temperature increase of 1·5 °C from 1990 to 2050) and a more extreme increase (temperature increase of 2·4 °C for the same time period). The two climate change assumptions roughly cover the range of likely climate change in Central Europe and correspond to the lower and the higher range of global circulation model runs for the IPCC A1B scenario (IPCC 2007). We do not account for possible spatial structure in temperature anomalies as the extent of the study area is rather small and such spatial structure is not predictable with sufficient precision at our fine spatial scale. We further suppose a linear temperature increase over the study period which is roughly in line with IPCC predictions up to the middle of the 21st century (IPCC 2007). This assumption results in a constant annual temperature increment of 0·025 °C and 0·04 °C for the two warming predictions, respectively. Furthermore, we assume that other features of the climate system (e.g. precipitation) will remain constant. For the more extreme predicted temperature increase, the ragweed habitat suitability of the main populated regions of Austria and Bavaria increases mainly in the range of 0·1–0·15 and in some parts up to 0·25 (Fig. 1a).

Figure 1.

Increase in habitat suitability H(x) for ragweed Ambrosia artemisiifolia L. in Austria and Bavaria under climate warming of 2·4 °C from 1990 to 2050. Shades of grey indicate the degree of increase (a black square indicates an increase in habitat suitability by 0·25) (a). Simulated pollen distribution (darkest shade indicates maximum pollen load) and positions of the pollen traps (black squares) numbered according to Table 2 (b). Population density in Austria and Bavaria. Shades indicate population density (a black square indicates a population density >200 people per km²) (c).

Pollen Load and Allergy Costs

Exposure to ragweed pollen may cause allergic reactions. Experiments have shown that average travel distances of ragweed pollen grains are rather low, with only 1% of the pollen being dispersed over distances larger than 1 km (Raynor, Ogden & Hayes 1970). On the other hand, under favourable weather conditions, substantial fractions of ragweed pollen can occasionally be transported over distances as large as hundreds of kilometres (long-distance dispersion) (Cecchi et al. 2007; Smith et al. 2008). However, Cecchi et al. (2010) have recently argued that pollen grains might lose their allergenic potential during long-distance travel in the upper atmosphere, where they are subjected to extreme conditions in terms of air temperature, humidity and solar radiation.

Until now, there is no method for unequivocally determining the origin of ragweed pollen detected in pollen traps. In the flat eastern regions of Eastern Austria, it is evident that long-range transport significantly contributes to ragweed pollen load, because the adjacent regions of Hungary and the Northern Balkans are already heavily infested with ragweed. In the more western regions of Austria and in Bavaria, which are separated from Hungary and the Balkans by distance and mountain ranges, ragweed pollen load is currently considerably lower and probably mostly of local origin.

For calculating pollen dispersal, we assumed that the drifting time for each pollen grain follows an exponential distribution with parameter τ and that the wind is of constant direction and velocity for each individual trajectory. We generated the dispersal kernel using averaged wind data from two weather stations (Wien Hohe Warte and Ried im Innkreis), smoothed in the angular coordinate, of August and September from 2000 to 2008 in 10-min intervals.

To determine the costs arising from ragweed allergies, we first calculated for each infested cell the impact of pollen dispersed from that cell to all other cells. We considered all infestations in the study area (i.e. Austria and Bavaria) and in those adjacent regions for which we had distribution data (Hungary, Slovenia, South Tyrol). To achieve an estimate of the spatial pollen distribution, we compared the annual pollen totals recorded at eight Austrian pollen-monitoring stations that operated volumetric spore traps of the Hirst design (Hirst 1952) (see Table 2) for four consecutive years with results of simulations. We determined τ by testing for the optimal agreement of recorded annual pollen sums and simulations, giving a value of 1·72 h (Fig. 1b).

Table 2. Annual ragweed pollen grain sums registered in eight Austrian pollen traps for the years 2008–2011. Missing data (n.d.) indicate traps temporarily out of function
 Trap 1Trap 2Trap 3Trap 4Trap 5Trap 6Trap 7Trap 8
200862n.d.747279032328862
2009n.d.670744213101792291178
2010692462256n.d.117362711
20113672468784n.d.741011329

As we lack information about the number of pollen grains produced in a single cell, we did not calculate the number of affected persons, but rather the total allergy costs per cell. We achieved this by scaling our result, as described below, to a value presented by Jäger (2006), who provided a first calculation of the annual allergy costs of ragweed in Austria (90 million € in 2005). In doing so, we multiplied for each grid cell the integrated pollen impact from all other cells, as calculated by the dispersion kernel, by human population size (Fig. 1c) in the focal cell (CIESIN 2005). We set the sum over all cells as proportional to the total allergy costs. For Austria and Bavaria together, we accordingly get allergy costs of 133 million € in 2005. Costs vary from about 35 € per person in those regions with widespread infestations (e.g. Eastern Austria) to negligible costs in higher mountainous regions. It has to be noted that these are effective numbers averaged over all persons irrespective whether susceptible to ragweed allergy or not, corresponding to 10 € per average person. This results in allergy costs of 200–250 € per allergic person, who represent 4–5% of the Austrian resident population (Jäger 2006; Burbach et al. 2009). This estimate includes indirect costs such as the loss of working days of allergenic persons and is close to the 303 € that Bullock et al. (2012) have calculated for the average European ragweed allergic person. We note that therapy costs for the relatively few extremely sensitive persons with asthmatic reactions can be much higher (Bullock et al. 2012).

We used the same methodology to calculate annual allergy costs for all our simulations until 2050. We did not apply discounting rates due to their severe inherent problems in long-term socio-economic analysis of global change issues (Stern 2007; Gardiner 2011). All costs were calculated in standardized current (as of 2012) costs, that is, we did not consider inflation to make current and future expenses directly comparable. In our simulations, costs to human health under the different climate and management scenarios are compared as mean annual allergy costs for the management period (2011–2050) and as allergy costs at the end of the management period (2050).

Management Strategies

The management strategies tested here consist of two components (survey and eradication), and their implementation follows the procedure presented by Richter et al. (2012). In brief, management costs are quantified in abstract units management effort units (MEU), whereby 0·1 MEU corresponds to the effort of surveying one cell and a value of 1 MEU to the effort necessary for successful eradication of existing populations from such a cell. The latter also includes subsequent monitoring of eradicated sites within a cell to suppress re-emergence from the seed bank. Under these assumptions, the most effective way to limit the spread of ragweed is to select grid cells for management in a fixed order based on the habitat suitability. Management therefore has to start in the most suitable cell and to end at a certain threshold. After reaching the threshold, management starts again in the cell with the highest habitat suitability. In the present work, the optimal threshold in terms of cost-effectiveness as detailed in Richter et al. (2012) was calculated for each assumed climatic trend (no temperature increase, moderate increase and more extreme temperature increase) and management effort, and the reported allergy costs are those corresponding to this optimal threshold.

Estimation of Management Costs

To calculate the costs of management strategies, we assumed that one person is capable of surveying a strip of 10 m width when searching for ragweed. For surveying 1 km², this results in 100 strips each with a length of 1 km. Assuming a walking speed of about 4 km h−1, it takes 25 h for surveying 1 km2 or about 100 days (with eight working hours per day) to survey one cell (~35 km2). With a gross wage of 25 € per hour including add-on costs, the absolute value for surveying one cell amounts to 20 000 €. With additional material expenses of 10 000 €, the costs for surveying add up to 30 000 € corresponding to 0·1 MEU. Therefore, the eradication and subsequent monitoring of one cell cost 0·3 million €, which corresponds to 1 MEU.

Results

Human Allergy Costs Under Current Climate

Our simulations show that the mean annual allergy costs will be 291 million € for the management period (2011–2050) (Fig. 2a). When applying a management regime that uses 50 MEUs (15 million € per year) annually, the mean allergy costs drop to 82 million € per year. When management effort is increased to 75, 100, 125 and 150 MEUs annually, allergy costs decrease to 45, 32, 26 and 22 million € per year, respectively. In contrast, in the business as usual scenario (without management), allergy costs will increase from 133 million € in 2005 to 422 million € in 2050 (Fig. 2b). With management, allergy costs in 2050 will drop to 27 and 14 million € when spending 50 and 75 MEUs, respectively, but do not significantly decline further when investing more MEUs.

Figure 2.

Mean annual allergy costs in Austria and Bavaria over the management period (2011–2050) (a). Allergy costs in the year 2050 for the three climate assumptions, both vs. management effort in MEU (b).

Human Allergy Costs Under Climate Change

Under climate change with an annual temperature increase of 0·025 and 0·04 °C, respectively, and without management, the projected mean annual allergy costs are between 333 and 365 million €, an increase of 15–25% compared with the results based on current climate (Fig. 2a). With an annual management investment of 50 MEUs (15 million € per year), this number will decrease to about 105 and 140 million € per year compared with 82 million € under the current climate. We found that spending ≥100 MEUs (corresponding to 30 million €) will only marginally decrease mean annual allergy costs compared with the model run without climate change.

Annual allergy costs at the end of the management period (2050) will rise to 522 and 596 million € for an annual temperature increase of 0·025 and 0·04 °C, respectively (Fig. 2b). Hence, in 2050 costs will be between 20 and 40% higher than under current climatic conditions. Investing 50 MEUs per year for management, that is, the budget that drastically reduces costs under no climate change, proved less successful when climate warming was accounted for: allergy costs will decrease down to 81 million € (moderate climate change) and 187 million € (more extreme climate change) until 2050, compared with 29 million € under current climate.

What is the Optimum Management Strategy?

Management of a given alien species has to balance the costs incurred through implementing a management programme against the forgone costs by reducing the negative impact of the target species. We found that under no or moderate climate change, there is a threshold at about 100 MEUs (30 million €) per year beyond which no additional reduction in health costs are achieved (Fig. 3). Under strong climate change, this threshold increases to about 125 MEUs (37·5 million €) per year. Accordingly, a cost-benefit analysis reveals that total savings from ragweed management over the whole period (2011–2050) amount to about 9 billion € under current climate, to about 11 billion € for moderate climate change and to about 12 billion € for more extreme climate change.

Figure 3.

Total management costs and allergy costs demonstrating the economic efficiency of different management strategies.

Discussion

Our analyses show that management of ragweed in Central Europe is profitable and therefore highly justifiable from an economic point of view. The costs of management stay clearly below the potential future costs, if spread of ragweed continues and is not reduced. This is true for constant climatic conditions as well as for climate change. Following an optimally allocated management strategy with an annual budget of 30 million € reduces mean allergy costs for no change in temperature and moderate (annual temperature increase of 0·025 °C) and more extreme (annual temperature increase of 0·04 °C) climate change by 258, 295 and 325 million € per year, respectively. Thus, management may yield substantial savings, in particular for the more extreme climate assumption, where total savings over forty years amount to about 12 billion €. Given the scale of impacts on human health, and the substantial gains accrued from management, our results suggest that it is wise to act early and determined to halt further spread of ragweed in Austria and Bavaria, and beyond.

Cost-Benefit Analyses and Biological Invasions

Our study illustrates that the management of invasive alien species has an economic benefit beside natural conservation using the example of ragweed in Austria and Bavaria. Previously, Shaw (1982) writing on weed management pointed out that ‘costs, benefits and risks must be carefully weighed before decisions can be made that are clearly in the public interest’. At this time, he was ‘confident that we will succeed’. Now, more than 30 years later, we have to be aware that decision-makers rarely take account of the value of scientific studies. As Joly et al. (2011) stated for ragweed at roadsides: ‘few municipal managers in charge of mowing the roadsides of local roads are aware of common ragweed or of protocols for its control’.

An important factor for our cost-benefit analyses is the allergy costs. The yearly health costs we have calculated for Austria and Bavaria (133 million € in 2005 for 21 million inhabitants) are in the same range as given by Ngom & Gosselin (2013) for the Canadian province of Québec (155 million CAD equal to about 110 million € for 8 million inhabitants). Together with the costs for management, we achieve a benefit–cost ratio of 10 : 1. This result is in line with McConnachie et al. (2003) who give a review of four terrestrial weeds with benefit–cost ratios from 2·3 : 1 to 20·7 : 1 and van Wilgen et al. (2004) who estimate the benefit–cost ratios from 8 : 1 to 709 : 1 and with other cost-benefit studies on alien plants, which have also shown that the gains of determined management outweigh the costs (e.g. Currie, Milton & Steenkamp 2009; Wise, Wilgen & Le Maitre 2012; Yokomizo et al. 2012). This is the case particularly if the focal species is not yet abundant, if per capita-costs caused are substantial and if efficient management measures are available. In Austria and Bavaria, where ragweed is already frequent in some parts and where its local eradication is relatively expensive, the high potential (health) costs are the main factor that renders management an economically viable investment.

In regions where ragweed has already become common as in south-eastern Europe (Bullock et al. 2012) and central North America (Ziska et al. 2011; Ngom & Gosselin 2013), management costs per unit area will be much higher, but so will be the benefits. In such heavily infested regions, complete eradication may be unrealistic, whereas a substantial reduction in population sizes still might be achievable.

Allergy and Management Costs Under Climate Warming

An annual warming of 0·025 or 0·04 °C as used in our calculations leads to considerable acceleration of the spread of ragweed. Even more alarmingly, management costs sufficient for halting ragweed's spread and even reducing the total area infected under current climatic conditions will have to be nearly doubled to curtail the spread of ragweed if the climate warms as predicted.

Future exposure of susceptible persons to ragweed pollen through spread of ragweed to currently uninfested areas will be a primary cause of increasing ragweed pollen allergies. We note, however, that our estimate of rising ragweed allergy costs under global warming might still be conservative because the period of pollen flight may also become prolonged under a warmer climate. Moreover, we have not taken into account potential savings of management regimes due to reduced agricultural losses caused by ragweed infestation of fields. If the latter are added, the economic value of a carefully designed ragweed management programme will probably be considerably higher.

Future Challenges

Cost-benefit analyses are increasingly conducted to weigh the costs and benefits of action and inaction against the spread of alien plant species (e.g. Currie, Milton & Steenkamp 2009; Frid et al.2013). Unavoidably, cost-benefit analyses of alien species management are subject to a range of simplifications and uncertainties (Keller, Lodge & Finnoff 2007; Sahlin et al. 2011; Yokomizo et al. 2012), which may affect optimal resource allocation and decision-making (e.g. Regan et al. 2005; Yokomizo et al. 2012). This also holds for the approach taken here. There are four fields which we consider particularly worthy of further improvement:

First, our model does currently not account for a natural background mortality rate and neglects the process of local population growth within infested cells. As a consequence, our calculations do not account for different population sizes and hence different eradication costs in cells colonized at different time points. Further, we assumed a perfect detection of ragweed in surveyed cells, and a perfect eradication success of managed populations.

Secondly, the values reported here are those that result from the optimally chosen management protocol. They imply that the species spread and pollen dispersal mechanisms are exactly known in a statistical sense, which is fulfilled in our simulations as the same models were used for management parametrization and evaluation. A real-world management implementation with the suitability threshold that is optimal for the simulation will thus perform worse than reported here. However, as Richter et al. (2012) have shown, the influence of the chosen threshold on the results is weak.

Thirdly, calculating the impact of ragweed on human health in financial terms is subject to a range of assumptions. These issues are related to methodology (e.g. which impacts are considered and how are they monetized), as well as to uncertainties on the scale of impacts (e.g. number of allergic people, which allergic reactions can be attributed to ragweed pollen) caused by ragweed. The proportion of allergic persons may rise with pollen concentration and with climate change. However, such a rise will even make a sound management strategy more rewarding.

Similarly, a recent study showed that increases in atmospheric concentrations of CO2 could lead to increased pollen production per plant (Ziello et al. 2011). Increased concentrations of atmospheric CO2 affect plant function (Ziska & Caulfield 2000), and so plants grown in higher CO2 environments generally grow faster, are larger at maturity and produce more pollen (Ziska & Caulfield 2000; Wayne et al. 2002). However, we did not consider this factor in our analysis because sound quantitative data for parameterizing this CO2 fertilization effect is not currently available.

Finally, the calculation of management costs is based on a crude estimate. We note that our estimates were purposefully set to rather high values. In particular, we assumed that ragweed might occupy any location within a given 3 × 5′ grid cell, although the habitat of considerable parts of many cells is unsuitable for ragweed (e.g. forests, grassland, water bodies). Consequently, the calculated optimal economic benefit of ragweed management is likely to be a conservative estimate and might actually be considerably higher.

Conclusions

Given the potential impacts of invasive plants, their control has become an important management issue, and there is a demand on applied ecological research delivering information to guide successful management programmes. However, despite considerable efforts on the part of the scientists, current research practice still seems to match the needs of managers rather poorly. Absent or incomplete cost-benefit analyses of management programmes are one prominent reason for this mismatch (Kettenring & Adams 2011) and are probably due to the inherent difficulties in valuing various effects of invasive plants, for example, on native biodiversity or ecosystem services (Hulme 2012). In comparison, impacts on human health are easier to quantify in economic terms, in particular in regions such as central Europe where the healthcare system involves a detailed documentation of cases, therapies and associated expenses. Despite a series of assumptions and problems, different efforts to calculate the allergy costs associated with ragweed spread have hence come to relatively similar results (Reinhardt et al. 2003; Jäger 2006; Bullock et al. 2012). As a corollary, providing sensible cost-benefit analyses to managers seems particularly feasible for weeds with human health impacts.

Based on such comparatively reliable cost estimates, the results of this study show that even an expensive long-term management programme for an already relatively widespread weed will definitely pay off in national economy terms. The magnitude of the benefit is such that various uncertainties that underlie our calculations of management costs could hardly affect our qualitative conclusions. Clearly, this result is not generalizable to all invasive species with health impacts, but it demonstrates that the benefits of management are potentially large. Similar studies on other weeds with economically valuable effects would hence be useful and might foster the public and political readiness to invest in longer-term management programmes.

Acknowledgements

This work was supported by the Austrian Academy of Sciences within the Global Change Programme, and the European Science Commission (FP 6 project 036866: ECOCHANGE). We are grateful to R. May, H. Niklfeld, L. Schratt-Ehrendorfer and T. Englisch for access to the data of the Floristic Mapping Projects of Austria and Germany. Valuable unpublished distribution data have been provided by numerous other colleagues. The results presented here relate to COST Action FA COST Action FA1203 Sustainable management of A. artemisiifolia in Europe (SMARTER). The comments and suggestions of Philip Hulme, Jennifer Firn and two referees helped to improve the article and are highly appreciated.

Ancillary