Keywords

1 Introduction

By ‘eradication’, invasion scientists mean removing every individual of a discrete, more or less isolated population. This is distinct from ‘extirpation’, which means eliminating a segment of a population, but with conspecific individuals still present in contiguous or nearby populations. Unfortunately, the term ‘eradication’ is used quite colloquially, particularly in media reports and political statements advocating or announcing ‘eradication’ of some weed or pest species, when what is meant is really extirpation of the species in some defined area. Sometimes ‘eradication’ is used to mean simply killing a lot of individuals, not even all of them at the same site. Part of a general scepticism about the feasibility of eradication stems from the fact that campaigns colloquially announced as ‘eradication’ campaigns were never meant to be that, so of course they failed to eradicate the target population (Simberloff 2003a).

Another factor generating scepticism about the possibility of eradication is the history of several high-profile, expensive true eradication campaigns that not only failed to eradicate their target species but had enormous damaging non-target impacts. An example is the failed campaign in the United States to eradicate white pine blister rust, introduced in the early twentieth century on white pine seedlings from Germany (Maloy 1997). The campaign aimed to eliminate the fungus by eradicating both native and introduced species of Ribes, the alternate host. Labour costs alone were over $150 million and were particularly heavy during World War I. During World War II, prison inmates as well as German and Italian prisoners of war dug up Ribes and spread chemicals, including in wetlands and stream-sides. Non-target impacts were massive, and the campaign failed utterly. Another continent-wide campaign, this time to eliminate Berberis vulgaris (European barberry), was similarly motivated. In the United States, B. vulgaris is an alternate host of stem rust of cereals, which inflicted enormous losses on wheat growers. The campaign, detailed by Campbell and Long (2001) and Mack and Foster (2009), began in 1918, lasted 60 years, employed thousands, and rendered B. vulgaris a rare plant in much of the United States, even today. As the methods included use of rock salt, kerosene, and dynamite (Mack and Foster 2009), one can speculate about non-target impacts.

In a widely cited paper, Rejmánek and Pitcairn (2002) found a decade ago that eradication of agricultural weed populations smaller than a hectare is usually feasible, and that for infestations between one hectare and 1,000 ha, between a fourth and a third of attempts they surveyed had succeeded. However, in their survey the cost of eradications rose so rapidly with area that they felt it was unlikely that eradications of plant populations occupying more than 1,000 ha would be feasible. Panetta and Timmins (2004) agreed with this threshold and suggested that the prospects for eradication of plants from natural areas would tend to be much dimmer than those for agricultural weeds. Gardener et al. (2010) recently cast further doubt on the feasibility of most plant eradication projects other than very small ones, an assertion subsequently publicised in a high-visibility news report (Vince 2011). However, the past decade has seen dramatic progress in eradication of invasive animal populations (see for example Genovesi 2011a, b) on ever-larger islands, plus further experience with invasive plant management. It is thus timely to reconsider Rejmánek and Pitcairn’s pessimism regarding large-scale plant eradications and also ask under what circumstances should attempted eradication be the preferred response to a plant invasion of a PA as opposed to some sort of maintenance management, such as biological, chemical, mechanical, or physical control.

It is a commonplace that plants are generally harder to eradicate than animals, especially vertebrates. Seed banks may persist in the soil for many years (Panetta 2004), numbers of plant individuals may be enormous, individuals – even seedlings of trees – may be small and cryptic, and the attractive baits and traps that have so aided animal eradication are not applicable to plants. Thus, the degree of optimism that has begun to infuse the community of managers and policymakers dealing with invasive animals (e.g. Genovesi 2011a, b) has largely failed to engage those managing invasive plants. It is telling that, in a recent international conference on eradication of invasive species on islands (Veitch et al. 2011), of 94 papers, 89 were about eradicating animals and only 6 about eradicating plants; of 45 abstracts, 44 were on eradicating animals and none were on eradicating plants.

2 Successes and Failures

Many small plant invasions have been eradicated from sites other than PAs. Mack and Lonsdale (2002) describe eradication in Australia of small populations of North American Eupatorium serotinum (late boneset) in a cattle sales yard, as well as eradication of Old World Centaurea trichocephala (feather-head knapweed) in a degraded pasture in Washington state, USA. In the marine realm, the “killer alga” (Caulerpa taxifolia) was eradicated from two sites in California (Anderson 2005; Woodfield and Merkel 2006). At one site, about 0.13 ha of the alga was distributed widely among 42.3 ha of a 100.6 ha lagoon, and the other consisted of a group of shallow ponds totalling 1.1 ha connected to a harbour. In South Australia, C. taxifolia was eradicated from an artificial marine water body 7 km long by a few hundred meters wide (Walters 2009).

Some small populations of invasive plants have been eradicated in PAs. For instance, Rejmánek and Pitcairn (2002) cite two eradications of small populations in the Channel Islands National Park, California, while Oryza rufipogon (Asian common rice) was eradicated from an area of 0.1 ha in Everglades National Park (Westbrooks 1993). Macdonald (1988) reports ten invasive plant species as having been eliminated from Kruger National Park, South Africa. He identified only four of these species: Opuntia aurantiaca (jointed cactus), Acacia dealbata (silver wattle), Bidens formosa (cosmos), and Nicotiana glauca (tree tobacco). However, L. Foxcroft (personal communication, 2013) reports that the latter species is present cyclically. The Bermuda Department of Agriculture has eradicated Livistona chinensis (Chinese fan palm), Pimenta dioica (allspice), Eugenia uniflora (Barbados cherry), and Citharexylum spinosum (fiddlewood) from Nonsuch Island (Bermuda), a wildlife sanctuary of 5.7 ha (Mack and Lonsdale 2002). In 1972, the New Zealand government targeted 29 non-native plant species for removal from 2,943 ha Raoul Island, a designated nature reserve. For seven species that occupied relatively small areas, including highly invasive ones such as Cortaderia selloana (pampas grass), success is believed to have been achieved, although continued monitoring is undertaken to ensure that resurgence does not occur from a soil seed bank (West 2002). However, for the seven main target species, all originally quite widespread on the island, progress toward eradication has been more gradual, with occasional setbacks as new infestations are detected (Holloran 2006). In the Galapagos, four non-native plant species have been eradicated from Santa Cruz Island (two of these are not found elsewhere in the archipelago), each from an area less than 0.1 ha (Gardener et al. 2010). Although the great majority of Santa Cruz is part of the Galapagos National Park, at least one of eradications took place on private land. Cenchrus echinatus (sandbur) was eradicated from 64 ha on Laysan Island (Hawaiian Islands; 411 ha), managed as a PA by the US Fish and Wildlife Service, in a 10-year campaign beginning in 1991 (Flint and Rehkemper 2002; E. Flint, personal communication, 2007).

A much larger success, although not in a PA, was the eradication of the pasture pest Bassia scoparia (burning bush) from several thousand ha distributed over a linear distance of 900 km in western Australia (Randall 2001; Dodd 2004), no doubt helped by the fact that locations of all plantings had been recorded. Perhaps the most ambitious current plant eradication programme rivals the Ribes and Berberis eradication campaigns of the early twentieth century. This is the attempt to eradicate a parasitic agricultural weed, Striga asiatica (witchweed), which is ongoing after over 50 years (Eplee 2001; Mack and Foster 2009) and has reduced the infested area from 162,000 ha to less than 1,000 ha in North and South Carolina. Success is likely within a decade (Mack and Foster 2009).

Many more attempted plant eradications have failed than have succeeded. Gardener et al. (2010), for example, cite failure to eradicate (so far) 26 targeted plant species in the Galapagos, comparing this record to the four successes cited above. For at least two of these failures, no campaign was actually implemented. For Caulerpa taxifolia, several eradication efforts have failed, as against the three successes noted above (Walters 2009).

3 Criteria for Success

Myers et al. (2000) and Simberloff (2002a, b, 2003a, b) have suggested several criteria that characterise successful eradications and that should be met before eradication is attempted. Of course the idiosyncrasies of each case will weigh heavily, but the following factors should always be borne in mind:

  1. 1.

    Economic resources. Are resources sufficient to complete the eradication as planned, and are those resources encumbered in such a way that they will be available for the duration of the project, even as the target population and its perceived impact are greatly reduced? Costs of removing the last few individuals may exceed those of removing all the rest, and funding agencies may be inclined to reduce support once the problem is lessened (Mack and Lonsdale 2002). Are resources needed to manage the species in areas near the target protected area to prevent reinvasion, and, if so, are they available?

  2. 2.

    Adequate lines of authority. Eradication is, by its nature, an all-or-none phenomenon. By contrast, in maintenance management by chemical or mechanical control, for instance, the refusal of a few landowners to permit the project to be carried out on their property would not necessarily prevent substantial reduction of the target species. However, an inviolable sanctuary for the target would prevent eradication by definition. Do such sanctuaries exist adjacent to or near the target protected area?

  3. 3.

    Enough must be known about the biology of the target species that a route to eradication can be identified that is feasible with available resources.

  4. 4.

    The eradication project, even if successful, must not produce an undesirable condition. For PAs, the ultimate goal would almost certainly be restoration of a semblance of the natural ecological community and the dynamic trajectory it was following before the invasion. Thus, for instance, high likelihood that an eradicated plant species would simply reinvade quickly or be replaced by another introduced species would weigh heavily against attempting eradication (although this would not necessarily be decisive; see an example below). It is also possible the method used in an attempted eradication would have a high risk of non-target impacts that would prevent the restoration goal from being achieved. Massive use of some persistent herbicide, for instance, or tremendous damage from machinery used in a scorched-earth operation, might so damage the prospects for restoration as to be untenable.

4 Monitoring: Determining Success and Detecting Reinvasion

The main issue concerning eradication of any invasive plant population is whether it is feasible and at what cost. If so, particularly if the campaign is costly, it then becomes important to consider whether reinvasion is likely, whether it would be detected quickly, and what could be done about it if it occurs. Protected areas adjacent to unprotected lands pose particular problems in this regards, whereas island refuges are obviously at an advantage. Of course to a great extent likelihood of reinvasion depends on the location of the site relative to extant populations and the means by which propagules of the eliminated species might arrive. Constant vigilance is needed first of all to ensure that an eradication effort really was successful, and secondly, to note and deal with any newly arrived individuals.

To know that every last individual of a plant species is gone is fraught with many difficulties, enumerated by Panetta and Timmins (2004). With plants, the existence of a soil seed bank poses particular problems (Panetta 2004) and, depending on seed longevity, can mean that many years must pass before one can ascertain that eradication had occurred. For animals, depending on the species, it is common practice to declare success (or concede failure) quite quickly, for example, often 1 year for rats and 4 years for the Asian longhorn beetle (Anoplophora glabripennis) in Chicago (Haack et al. 2010). For plants, sometimes 4 years of absence has been chosen as the criterion for success, for example, C. taxifolia in California (Woodfield and Merkel 2006), while for Bassia scoparia in Western Australia the criterion was 3 years (Randall 2001). However, several announcements of eradication have been premature. For instance, in Queensland, a 40-plus-year campaign to eradicate several small populations of the North American herb Helenium amarum (yellow sneezeweed), first detected in 1953, was declared successful in 2002 after annual searching for survivors failed to detect any (Mack and Lonsdale 2002; Csurhes and Zhou 2008). But in 2007 several individuals were found; these were removed and the area continues to be monitored (Csurhes and Zhou 2008).

5 When Should Eradication Be Attempted in Protected Areas?

Not all potential eradication projects that meet the above criteria for high likelihood of success can be undertaken, if only because resources would likely not suffice. However, in assessment of alternative management possibilities – in essence, (i) do nothing, at least for the present, (ii) attempt some sort of maintenance management, or (iii) attempt eradication – several factors suggest that eradication deserves more consideration than it often gets.

First, if an invasion is recent and the invaded area still small, it is likely that eradication is feasible, as suggested by the data in Rejmánek and Pitcairn (2002; see Pluess et al. 2012). Furthermore, the cost would be far less than if the effort were made after the invasion had spread. The likelihood of damaging non-target impacts would be less, both because the invader is unlikely to have established important interactions with native resident species and because whatever eradication method is attempted will not be employed over a large area. Finally, it is risky to wait to see if the species begins to spread or cause problems and, if it does, only then undertake an eradication campaign. Many introduced species, including plants such as Schinus terebinthifolius (Brazilian pepper) and Arundo donax (giant reed), have remained restricted for long periods, even decades, before rather suddenly spreading widely (see Crooks 2005). Because eradication campaigns typically take time to plan and implement, one could easily miss a window of opportunity by delaying an eradication attempt. Also, some invasive plants have major ecosystemic impacts that are nevertheless sufficiently subtle that they are not detected quickly; plants that fix nitrogen or concentrate phosphorus can fall in this category (Simberloff 2011). Waiting until such impacts become evident may allow a species to spread to a point at which eradication is vastly more expensive and perhaps not feasible. Crupina vulgaris (common crupina) in the American West and Clidemia hirta (Koster’s curse) in the Hawaiian Islands, two non-native plants that were discovered soon after arrival and almost certainly could have been eradicated, without likely reinvasion, were allowed to spread while authorities questioned whether they would be very damaging. Both proved highly invasive and were well beyond the stage when they could have been eradicated by the time it was agreed that they should be controlled (Simberloff 2003b). The alga Caulerpa taxifolia, which has now spread throughout much of the near-shore western Mediterranean, could also almost certainly have been controlled had a campaign been undertaken soon after discovery (Meinesz 1999).

For PAs, the status of the target plant in neighbouring areas is a particular concern, as the funding for management, including attempted eradication, in the PA is unlikely to allow efforts beyond that area. Thus, for instance, the State of Florida and US federal agencies have mounted a promising programme using chemical and mechanical means to reduce or eliminate Melaleuca quinquenervia (broad-leaved paperbark tree) from state and federal lands (including PAs) in south Florida (F. Laroche, personal communication). However, by statute public funds cannot support such efforts on private lands adjacent to government properties. Three biological control insects have been released and, of course, do not respect property boundaries. These may contribute to an effective maintenance management programme in this case, but biological control in otherwise untreated areas would be unlikely to lead to eradication of this or other invasive plant species if this were the goal.

For widespread invasions, including longstanding ones, the expense of an eradication campaign can be forbidding even if the technology exists to suggest that success is possible. However, a comparison to ongoing costs of maintenance management in some cases leads to speculation about whether attempting eradication might be the truly most cost-effective approach (Simberloff 2003a). For instance, the United States spends $45 million annually on management of Lythrum salicaria (purple loosestrife) and $3 million to $6 million annually on control of M. quinquenervia). Thus, over a 10-year period, ongoing maintenance management costs tens or hundreds of millions of dollars. One can imagine that having such resources available over a much shorter period for an eradication attempt might make an eradication attempt feasible that would have been impossible with just a few million dollars. Another possible resource that has not been devoted to plant eradication is volunteer or prisoner labour. Such sources are now routinely used in a number of effective maintenance management programmes, especially in PAs (see Simberloff 2003a), and allow managers to marshal many more workers than could possibly have been paid. Use of vast amounts of manpower might make it possible to eradicate much more widespread invasions than would have been deemed feasible based on personnel costs alone.

If an eradication campaign uses the same method that would have been used had maintenance management been the goal, it may well be more cost-effective to invest added resources and attempt to eradicate the invader. This is because even failure to eradicate would be no great loss and would probably entail more complete maintenance management. One would have to tally the costs and potential benefits of the added effort. An excellent example is the project, begun in 1992, to eradicate Ammophila arenaria (European beachgrass) and hybrids of two African ice-plant species (Carpobrotus edulis × C. chilensis) from an 11 ha area of Lanphere Dunes in Humboldt Bay National Wildlife Refuge in California (Pickart 2013). The stated goal in terms of these invasive plants was eradication, and the goal for the system was restoration of the ecosystem to its trajectory before European modification, by restoring abiotic processes that maintain a dynamic dune ecosystem. Herbicides were precluded by local community objections, and the impact of heavy machinery on native vegetation, including two federally listed species, would have been too great, so the method chosen was digging and pulling by hand. Ammophila arenaria was almost wholly eliminated after 2 years, and ice-plant after 5. However, rare resprouts of both species are seen, and there is occasional reinvasion from nearby areas. Annual spot treatments control these at very low densities. Thus, complete eradication has not yet been achieved for either species (or else reinvasion quickly occurs), but the ultimate restoration goal has been met, and, if maintenance management rather than total eradication had been the stated goal, the method that was implemented would have been exactly the same. Further, the fact that quick reinvasion is likely does not invalidate the approach in this instance, as annual monitoring and spot treatments are feasible and inexpensive.

6 Further Advances?

Just within the last decade, animal eradication has advanced greatly, with projects that would have seemed impossible a decade or two ago now well within the realm of possibility (Genovesi 2011a, b), with better methods of avoiding non-target impacts (e.g. Caut et al. 2009) and with important conservation benefits (McGeoch et al. 2010). Some of these advances result from new technologies and others from incremental improvement of existing techniques, combined with ambition (Simberloff et al. 2013). Even if one grants the difficulties that are peculiar to invasive plant eradication, it seems as if greater successes are possible by the same routes that are forging progress in animal eradication. For instance, greater efforts using the same techniques that had previously failed to control invasive plant species on Motuopao Island (New Zealand) are leading towards successful eradication of several species (Beauchamp and Ward 2011). Assiduous application of longstanding techniques has led to eradication of small infestations of 12 non-native plant species from single islands in the Hawaiian archipelago and the imminent elimination of eight others (Penniman et al. 2011).

What do not seem to have arisen in invasive plant control generally, and eradication attempts in particular, are highly innovative new technologies. Meyer et al. (2011) suggest a new strategy, focused on preventing fruit production, which might permit eradication of small infestations of previously intractable Miconia calvescens (miconia). But absent are plant analogues to completely novel approaches such as the development of attractive pheromones that have greatly advanced sea lamprey management (Fine and Sorensen 2008), toxic micro-beads that have cleared some water facilities of zebra mussels (Aldridge et al. 2006), and the battery of genetic manipulations currently under way in attempts to eradicate populations of fishes (e.g. Thresher 2008) and insects (e.g. Pollack 2011). It seems unlikely that the biology of plants differs in characteristic ways from that of animals so as to inhibit the development of radically new control technologies.