تجزیه و تحلیل اقتصادی فاصله ای از تشخیص و استراتژی های واکنش سریع برای گونه تهاجمی

Economic impacts from invasive species, conveyed as expected damages to assets from invasion and expected costs of successful prevention and/or removal, may vary significantly across spatially differentiated landscapes. We develop a spatial–dynamic model for optimal early detection and rapid response (EDRR) policies, commonly exploited in the management of potential invaders around the world, and apply it to the case of the Brown treesnake (Boiga irregularis) in Oahu, Hawaii. EDRR consists of search activities beyond the ports of entry, where search (and potentially removal) efforts are targeted toward areas where credible evidence suggests the presence of an invader. EDRR costs are a spatially dependent variable related to the ease or difficulty of searching an area, while damages are assumed to be a population-dependent variable. A myopic strategy in which search only occurs when and where current expected net returns are positive is attractive to managers, and, we find, significantly lowers present value losses (by $270 m over 30 years). We find further that in the tradeoff between search costs and damages avoided, early and aggressive measures that search some high priority areas beyond points of entry even when current costs of search exceed current damages can save the island more ($295 m over 30 years). Extensive or non-targeted search is not advised however.

Management of invasive species presents spatial and temporal analytical challenges that require integrated biological and economic modeling. An invasive species may be intercepted or treated across the invasion timeline: before entry to a new location (prevention), shortly after introduction but before establishment (early detection and rapid response, EDRR), by restricting it to a location smaller than the potential host range after establishment (containment), by restricting a population through harvest (control), or by allowing it to become part of the ecosystem (adaptation). Clearly, there are certain complementary and substitutable activities across this spectrum, but hitherto, applied research integrating optimal management of invasive species generally has not accounted for EDRR or spatial variation. This has kept the focus clearly on the important intertemporal tradeoffs in invasive species management (Burnett et al., 2006, Burnett et al., 2008, Finnoff et al., 2007 and Olson and Roy, 2005). However the addition of a spatial dimension has been shown to change optimal policy for control efforts when marginal costs of control include a cost of search, increasing the steady state level of a controlled invader as the spatial unit of analysis decreases (Burnett and Kaiser, 2007), and therefore must be explicitly included in any EDRR analysis. In this paper, we exploit the significant biological and economic research to date on the potential ecological and economic damages and costs of a particularly well-studied species of significant concern, the Brown treesnake (Boiga irregularis). 1 The Brown treesnake's imminent arrival in Hawaii ( Rodda et al., 1992, Rodda et al., 1999, Burnett et al., 2006 and Shwiff et al., 2010) provides an excellent case study to develop a spatially explicit, comprehensive dynamic EDRR management strategy to minimize the expected impacts of a potential invader. We contribute to the current literature in several respects, with a chief goal being to evaluate real-world invasive species management decisions in a bioeconomic framework. First, we consider EDRR, a real-world policy instrument commonly exploited in the management of potential invaders around the world, although not explicitly analyzed as a policy option in the literature to date. Second, we attempt to mimic real decisions facing managers with long-run dynamic consequences by examining decisions made across brief time horizons and assessing the impact of this constraint. Finally, our work expands, using real world data, the findings of Finnoff et al. (2007) that managers should prefer prevention to control even when their risk preferences lead them to wait for an invasion before treating. To do this, we ask whether a myopic policy under which only locations where the expectation is that current benefits outweigh current costs are searched is preferable to a strategy where more aggressive EDRR occurs so that certain locations are searched even when current net benefits may be negative. EDRR, defined here as intervention that occurs shortly after introduction but before there is a known population in a new location, consists of search activities beyond the ports of entry, where search (and potentially removal) efforts are targeted toward areas where credible evidence suggests the presence of an invader. EDRR should not be simply considered either ex-post prevention2 or low-population control (though both are components of EDRR) and deserves much greater analytical attention. This is due to the need to make decisions based on the possibility that a specimen is present across many possible locations. Our paper is a first step in formally modeling EDRR as an invasive species management tool. In order to concentrate on the combination of spatial and temporal components and the comparison of real-world myopic policy to optimal policy, we set aside virtually all uncertainty, using a deterministic representation of expected outcomes of a new invasion based on estimated population growth, costs of search treatments, and damages. As such, we can concentrate on the benefit EDRR adds to the management strategies of prevention and/or control, given a new invasion. Inspection, barriers around ports of entry, or any other action taken to avoid invasion is typically thought of as prevention. Prevention differs from EDRR particularly as the opportunities for reaping high returns are foregone once a species has successfully passed any bottleneck entry conditions where intervention could occur. After a new species has arrived and is established, attempts to reduce the new population can be categorized as control. Control differs from EDRR particularly as control can be considered harvest of an unwanted species and planning can compare population-dependent harvest costs with population-dependent damages. When discussing EDRR, the time frame is generally considered the time between when a species is likely to have first arrived based on knowledge of entry vectors and the time elapsed from that point until the new species’ presence should be noticeable without directed search. For the Brown treesnake, we estimate this to be about 30 years, which is a relatively long planning horizon for land managers to consider. Currently EDRR is a response driven policy tool that is myopic, in particular in the sense that it is not applied based on the very real likelihood that prevention has at some point failed (Burnett et al., 2008), but rather when a credible sighting of the species occurs. It has also been rather haphazardly funded. To analyze the implications of this on decision-making, we examine the impact of restricted planning horizons, under which managers cannot expect funds to be available over the entire 30-year period. Previous work focusing on long-run intertemporal tradeoffs has rightly been criticized for its lack of a spatial dimension, thus by including shorter time horizons we seek to identify how the time frame influences actual EDRR decisions (and avoid the reverse pitfall).

We develop a spatial dynamic model and solve for the optimal search strategy for a location needing EDRR. Such cases exist when prevention has a high likelihood of failing and/or incipient populations are very difficult to detect; these are common conditions for invaders. We then apply this model to the case of the Brown treesnake in Hawaii. We find that effective EDRR search targets limited areas of high expected net damages. Only 10% of the island needs treatment in a 30-year period, if it is applied efficiently, with 1.3% of the island receiving more than one treatment. A myopic strategy where cells are only treated when current benefits (damages foregone) exceed current costs also significantly reduces social welfare losses. In comparing the optimal policy to the myopic policy, we see that more aggressive, yet carefully targeted, search is preferred to a point, but that aggressive policies that are random in nature are more costly than either the optimal or the myopic strategies. Our application technique also allows examination of several simpler strategies that policy managers might adapt, given temporal and/or financial budgets, and identifies where search should be currently occurring, across the landscape, for maximum benefit under these conditions. A significant debate regarding EDRR hinges on whether overall spending is too little or too much – too little if the budget is spread too thin to do any permanent good; too much if we are searching for highly unlikely snakes now. Our results indicate that under virtually all expected conditions, any targeted search in high impact areas (where damages from the species would be relatively high and search costs are relatively low) will increase social welfare by reducing damages from the snake.19 This strong result should encourage Hawaii and other locations threatened by such stealthy invaders to increase EDRR efforts strategically, mapping out expected net damages and engaging in EDRR search in high impact areas, which may differ from ports of entry where prevention activities are focused. Spatial analysis using Geographical Information Systems (GIS) software and integrating biological parameters with economic ones can assist in developing optimal prevention and EDRR policies for invasive species as layers of information regarding damages, costs, and biological growth can be coalesced for analysis. This result also should strengthen the generalized debate, ranging from climate change to prevention screening of all sorts, for example, over whether early action, when environmental threats may seem small or have low expected value, is worthwhile. The spatial component is particularly relevant to the case of wildfire monitoring, where early fires, detected in remote areas, are expensive to detect but can be extinguished rather easily, while once the fire is easily visible (say, by satellite imaging), it is often too late to intervene and damages are much greater than they could be. Currently, no known snake populations exist on Oahu, but there is general agreement amongst the scientific community that there may be a small population. We begin our analysis with n0 = 1. 20 Current search on Oahu occurs only after a suspected sighting, while all other funds are expended on Guam and are targeted at preventing snake arrival at defined points of entry. Previous research ( Burnett et al., 2006) indicates that this may actually focus too much on the points of entry if snakes have already evaded detection there. Our results concur. Inefficient search, on the other hand, can be extremely costly, if it is random or incomplete. However comprehensive island-wide searches can reduce social welfare damages and may have additional external benefits, especially if prevention at entry points is highly effective at reducing the hazard rate. While the new evidence compiled for this study presents a myriad of opportunities for conclusions and controversy, we focus on four main findings we believe will stand the test of time and that extend beyond the case study presented here. First, treating EDRR as a separate but vital link between prevention at points of entry and control of known populations allows for insights into the costs of delay at low invasion population levels. In particular, in the absence of sufficient funding to assure eradication, within 5 years of an invasion it becomes preferable to pursue limited EDRR, in most cases even if it is not optimally allocated, as compared to doing nothing. This is true even though there are no areas after only 5 years where current damages exceed current costs of search – in fact there are no such areas until after 12 years. The benefits of imperfectly administering EDRR increase for the first 20–25 years, and then begin to decrease as damages accumulate. Given the handful of captures and the large number of credible snake sightings on Oahu, these results inform us that if the snake did in fact arrive on island even as far back as the 1980s, it is not too late to prevent significant losses in the future by acting today, though the window of opportunity is rapidly shrinking. Second, getting ahead of the problem pays off. The pattern for optimal targeted search is more aggressive than a myopic strategy that delays until current damages outweigh current costs. Furthermore, limited targeted search based on bioeconomic considerations of costs, damages, and growth (covering perhaps 10% of the island) outperforms random search significantly. Third, in spite of the fact that eradication through concerted island-wide sweeps can be profitable, it is never optimal in our model. Thus, though managers often tout eradication as the best management policy, few economic studies to date, including this one, have been able to verify this. While economic models have proposed eradication as a possible corner solution (Kaiser et al., 2007 and Olson and Roy, 2005), few case studies have found that it is preferred over some sort of interior solution. With budget constraints and myopic decision-making opportunities, it is beneficial to know that positive and significant returns can be generated by sporadic and incomplete treatments, especially if they are targeted to areas of high net expected damages. The threat of re-invasion reduces the advantages of eradication, but still eradication could be preferred to unmitigated growth. Finally, incomplete treatments should not be simple extensions of prevention that focus efforts solely on areas adjacent to points of entry. The simple analytical tools and widely available data used here can be tuned to reduce search costs and increase the reduction in damages. In the case of the Brown treesnake, EDRR should be applied to high population density areas as well as areas that serve both as conduits to new territory (roads) and areas that would experience particularly high damages from high snake populations in them. Certainly areas closer to points of entry are likely to have higher invasion populations, however optimal search efforts will also weigh the net expected damages to locate the most efficient search locations and times.